<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Inclusion Cloud  on Medium]]></title>
        <description><![CDATA[Stories by Inclusion Cloud  on Medium]]></description>
        <link>https://medium.com/@inclusion.cloud?source=rss-aa686e488265------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*Stwmmt1b-ahqQxNQE4P1Vw.png</url>
            <title>Stories by Inclusion Cloud  on Medium</title>
            <link>https://medium.com/@inclusion.cloud?source=rss-aa686e488265------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 05:59:47 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@inclusion.cloud/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[SAP MM in S/4HANA: Is It Worth Migrating from ECC?]]></title>
            <link>https://medium.com/@inclusion.cloud/sap-mm-in-s-4hana-is-it-worth-migrating-from-ecc-f6a6cea7b91e?source=rss-aa686e488265------2</link>
            <guid isPermaLink="false">https://medium.com/p/f6a6cea7b91e</guid>
            <category><![CDATA[materials-management]]></category>
            <category><![CDATA[sap-s4hana]]></category>
            <category><![CDATA[sap]]></category>
            <category><![CDATA[supply-chain]]></category>
            <category><![CDATA[erp-software]]></category>
            <dc:creator><![CDATA[Inclusion Cloud ]]></dc:creator>
            <pubDate>Fri, 13 Nov 2026 13:40:25 GMT</pubDate>
            <atom:updated>2026-11-13T13:40:25.142Z</atom:updated>
            <content:encoded><![CDATA[<p>We’ve previously discussed <a href="https://inclusioncloud.com/insights/blog/modernization-sap-migration/">when it makes sense to modernize during an SAP migration</a> and how to avoid adding unnecessary complexity to the process.</p><p>This time, we’re focusing on one specific part of the ERP: <strong>SAP Materials Management (MM)</strong>.</p><p>For manufacturers in industries like automotive, food and beverage, and chemicals, MM is involved in everyday operations. From ordering raw materials and managing suppliers to tracking inventory and recording deliveries, it helps keep production moving.</p><p>And while these processes are already supported in SAP ECC, SAP introduced several new capabilities in 2026 that are worth looking at if you’re considering a move to S/4HANA.</p><p>Here are five updates, what they actually do, and how they could make life easier for procurement and supply chain teams.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*OekLFya8G1PFrtu7.png" /></figure><h3>1. Update Multiple Purchase Order Delivery Dates with Joule</h3><p><strong>Released:</strong> February 2026 | SAP S/4HANA Cloud Public Edition 2602</p><p>A supplier calls to say several deliveries will arrive a week late. Now someone from procurement has to find the affected purchase orders and update their delivery dates.</p><p>It’s a routine task, but it can take quite a bit of time when you’re dealing with dozens of orders.</p><p>With the new Joule capability, buyers can request these changes <strong>using natural language</strong>. <a href="https://inclusioncloud.com/insights/control-joule-ai-unit-spending-sap/">Joule</a> identifies the relevant purchase order items, schedules the updates, and lets users monitor the results.</p><p>Instead of opening each order individually, a buyer can ask Joule to update delivery dates across multiple orders.</p><p><strong>Why it matters:</strong> Buyers spend less time updating purchase orders and more time managing suppliers and potential delays. Keeping delivery dates current also helps production and warehouse teams plan around when materials are expected to arrive.</p><p><strong>Available in ECC?</strong> No. ECC supports purchase order changes, including mass-processing options, but this Joule capability is specific to <a href="https://community.sap.com/t5/enterprise-resource-planning-blog-posts-by-sap/sourcing-amp-procurement-in-sap-cloud-erp-2602/ba-p/14307481">S/4HANA Cloud Public Edition</a>.</p><h3>2. Create Multiple Supplier Confirmations in One Step</h3><p><strong>Released:</strong> February 2026 | SAP S/4HANA Cloud Public Edition 2602</p><p>Once a purchase order is sent, procurement still needs to know whether the supplier can deliver the requested materials on time.</p><p>That’s where supplier confirmations come in. They help buyers keep track of what suppliers have committed to deliver and when.</p><p>SAP introduced a Joule capability that allows buyers to create confirmations for <strong>up to 10 purchase orders at once</strong> through a single conversation.</p><p>Think of an automotive manufacturer receiving confirmations for components from several suppliers. Instead of processing each eligible order separately, the purchasing team can handle multiple confirmations together.</p><p><strong>Why it matters:</strong> Less time entering confirmations means more time checking whether suppliers can meet delivery commitments, especially when production depends on materials arriving on schedule.</p><p><strong>Available in ECC?</strong> No. ECC already supports supplier confirmations, but this Joule-based mass creation capability was introduced for S/4HANA Cloud Public Edition.</p><h3>3. Confirm Goods Receipts Through a Conversation</h3><p><strong>Released:</strong> August 2026 | SAP S/4HANA Cloud Public Edition 2608</p><p>Ordering materials is only half the process. Once they arrive, someone needs to confirm that the goods were received so the ERP reflects what actually happened.</p><p>SAP’s new <strong>Goods and Services Receipt Agent</strong> makes this easier for employees who occasionally need to confirm deliveries.</p><p>For example, someone in a manufacturing plant orders maintenance supplies through a self-service purchase requisition. When the items arrive, they can use Joule to confirm the receipt through a conversation, rather than navigating a separate SAP application.</p><p>The agent guides them through the process and creates the corresponding record.</p><p><strong>Why it matters:</strong> Employees who don’t work in procurement every day can confirm receipts more easily. That helps purchasing teams keep records up to date and reduces the need to follow up on deliveries that have already arrived.</p><p><strong>Available in ECC?</strong> No. ECC supports goods receipts through transactions such as MIGO, but this <a href="https://community.sap.com/t5/enterprise-resource-planning-blog-posts-by-sap/sourcing-amp-procurement-in-sap-cloud-erp-2608/ba-p/14445515">conversational capability</a> is part of S/4HANA Cloud Public Edition.</p><h3>4. Update Custom Fields Across Multiple Purchase Orders</h3><p><strong>Released:</strong> August 2026 | SAP S/4HANA Cloud Public Edition 2608</p><p>Not every improvement needs AI to make a difference.</p><p>Procurement teams often use custom fields to track information specific to their business, such as internal project references or purchasing classifications.</p><p>The problem comes when that information needs to change across a large number of purchase orders.</p><p>With the updated <strong>Mass Changes to Purchase Orders</strong> application, buyers can modify custom fields across multiple purchase order items, even when they belong to different orders.</p><p>Say a manufacturer needs to change an internal classification across 100 purchase order lines. Instead of editing each one, the team can apply the change in a single mass-processing operation.</p><p><strong>Why it matters:</strong> This is especially useful for purchasing teams handling large order volumes. It cuts down on repetitive data entry and helps keep information consistent across orders.</p><p><strong>Available in ECC?</strong> Not as this standard Fiori enhancement. ECC offers mass-change capabilities and custom solutions, but this <a href="https://community.sap.com/t5/enterprise-resource-planning-blog-posts-by-sap/sourcing-amp-procurement-in-sap-cloud-erp-2608/ba-p/14445515">specific functionality</a> was introduced in S/4HANA Cloud Public Edition 2608.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Do47jdiicaSTf4LV.png" /></figure><h3>5. Find and Track Material Movements More Easily</h3><p><strong>Released:</strong> August 2026 | SAP S/4HANA Cloud Public Edition 2608</p><p>Anyone working in inventory management knows how frustrating it can be when the numbers don’t match.</p><p>Maybe a delivery was received but the stock isn’t showing where expected. Or a shipment was recorded, but someone needs to trace the transaction.</p><p>Finding the right material document is often the first step in understanding what happened.</p><p>SAP enhanced the <strong>Material Documents Overview</strong> application with additional search fields for inbound and outbound deliveries, customer returns, and goods receipt or issue slip numbers.</p><p>This means a warehouse supervisor investigating a stock discrepancy can search using a delivery reference and more easily locate the associated material documents.</p><p><strong>Why it matters:</strong> Inventory and warehouse teams can spend less time searching through transactions and more time resolving discrepancies, checking movements, and keeping operations running.</p><p><strong>Available in ECC?</strong> Not as this Fiori enhancement. ECC already provides material document reporting through transactions such as MB51, but these <a href="https://help.sap.com/docs/PRODUCTS/ee9ee0ca4c3942068ea584d2f929b5b1/caac39d1c20741a08f4d2d995610b68f.html">additional search capabilities</a> belong to S/4HANA Cloud Public Edition 2608.</p><h3>What Do These Updates Mean for Companies Still Running SAP ECC?</h3><p>Looking at these updates, it’s pretty clear that SAP is trying to address some of the everyday pain points that have traditionally required a lot of manual work or left room for errors.</p><p>Think about how much time procurement and supply chain teams spend updating purchase orders, confirming deliveries, or tracking materials. These might seem like small tasks, but they add up, especially in companies managing thousands of materials and transactions.</p><p>And we know how <strong>expensive mistakes</strong> can be in this area. An incorrect purchase order or an inventory discrepancy can affect production schedules, delay deliveries, and create problems across the entire supply chain.</p><p>That’s why we see real potential in making these processes easier to manage. Automating repetitive tasks can free up time for more important work, while conversational AI could make SAP easier to navigate and reduce the learning curve for business users.</p><p>There’s some interesting research supporting this potential. A <a href="https://academic.oup.com/qje/article/140/2/889/7990658">study published in <em>The Quarterly Journal of Economics</em></a> found that employees using a generative AI assistant were <strong>15% more productive on average</strong>, with even greater gains among less experienced workers.</p><p>Of course, deciding whether to migrate to S/4HANA involves much more than these new features. There’s the investment, potential disruption to operations, organizational readiness, and the resources needed to handle a complex migration.</p><p>At <a href="https://inclusioncloud.com/technologies/sap-consulting-and-solutions/">Inclusion Cloud</a>, we’re an SAP partner with certified resources experienced in complex migrations and integrations. If your team is considering a move to S/4HANA, we’d be happy to share more about our delivery capabilities and how we could support your project.</p><p><a href="https://inclusioncloud.com/contact-us/">Book a call with our specialists.</a></p><p><em>Originally published at </em><a href="https://inclusioncloud.com/insights/blog/sap-mm-s4hana-new-features/"><em>https://inclusioncloud.com</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f6a6cea7b91e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Agentic Software Delivery: Compact Teams at Greater Scale]]></title>
            <link>https://medium.com/@inclusion.cloud/agentic-software-delivery-compact-teams-at-greater-scale-569f2a708319?source=rss-aa686e488265------2</link>
            <guid isPermaLink="false">https://medium.com/p/569f2a708319</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[ai-agent]]></category>
            <dc:creator><![CDATA[Inclusion Cloud ]]></dc:creator>
            <pubDate>Thu, 05 Nov 2026 14:10:43 GMT</pubDate>
            <atom:updated>2026-11-05T14:10:43.745Z</atom:updated>
            <content:encoded><![CDATA[<p>At the start of the workday, a software engineering lead opens the project dashboard and finds more than a status update.</p><p>A new feature is ready for review. Its requirements have been refined, the initial implementation is complete, and the first round of tests has already surfaced a few edge cases. Dependencies have been checked, and the decisions still requiring human input are clearly outlined.</p><p>No one stayed late. No overtime was needed. <strong>AI agents</strong> had been working through the night.</p><p>This is one of the biggest changes <strong>agentic software delivery</strong> brings to software development. Agents can work through structured tasks around the clock, while experienced engineers set direction, supervise their output, and make the decisions that require human judgment.</p><p>The result? A <strong>compact team</strong> can <strong>take on</strong> <strong>more work </strong>and<strong> move faster</strong> without having to scale headcount at the same pace as its software roadmap. But there is another important shift.</p><p>When agents can access a <strong>shared knowledge layer</strong> containing project requirements, architecture decisions, coding standards, and business rules, they spend less time waiting for information or passing work between teams. They can <strong>use that context to carry tasks</strong> <strong>forward</strong> with greater autonomy.</p><p>This is what makes an <strong>AI-native approach to software delivery</strong> so promising.</p><h3>Automating the Software Development Lifecycle</h3><p><strong>Software development </strong>has traditionally moved through a <strong>series of stages:</strong> requirements gathering, solution design, development, testing, integration, and deployment.</p><p>Each stage produces work that someone else must review, interpret, and pass along. Even when organizations automate parts of the process, people still spend considerable time clarifying requirements, resolving inconsistencies, and coordinating work between teams.</p><p><strong>Agentic AI</strong> can help change that.</p><p>Traditional <strong>continuous integration and continuous delivery</strong> (CI/CD) pipelines have automated important parts of testing, integration, and deployment. Yet, according to a <a href="https://www.mckinsey.com/capabilities/mckinsey-technology/our-insights/rewiring-software-delivery-for-the-agentic-era#/">McKinsey report</a>, these activities can <strong>account for as much as 30% of total technology spend.</strong> Much of the work that comes before them, from defining requirements to writing code, remains manual and interpretation-heavy.</p><p>This is where <strong>agentic software delivery</strong> can make a difference. AI agents can help refine requirements, translate specifications into development tasks, check architectural constraints, generate code, run tests, and prepare results for review.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*wnNbeh9pg2LdN9QX.jpeg" /></figure><p>When these activities are connected through <strong>standardized workflows</strong> and <strong>machine-readable artifacts</strong>, agents can move work from one stage to the next with fewer manual handoffs. Engineers can spend less time coordinating tasks and more time <strong>reviewing what matters.</strong></p><p>In fact, the same McKinsey report signals that companies redesigning their software delivery models around agentic AI are seeing <strong>productivity improvements of 3x to 5x.</strong> These figures reflect reported experiences, not a guaranteed result for every organization. Still, they point to the potential of changing how work gets done across the lifecycle.</p><p><a href="https://inclusioncloud.com/insights/blog/the-exception-queue-could-be-killing-your-agent-roi/"><strong>Human oversight</strong></a> remains essential. <strong>Engineers</strong> and <strong>product leaders</strong> need to review critical outputs, validate quality, and ensure that changes meet business and technical requirements. The goal is to <strong>automate the handoffs between development stages</strong>, not just the tasks within them.</p><p>In short, agents can carry work from requirements to tested code, while engineers focus on architecture, validation, and decisions that require human judgment.</p><h3>Building the Knowledge Layer Behind AI-Native Software Delivery</h3><p>Now, as we have heard many times before, <strong>AI agents need to understand the environment</strong> in which they are working to actually perform tasks autonomously. Essentially, that means knowing <strong>four things:</strong></p><ul><li>How applications are structured.</li><li>Where enterprise data resides.</li><li>Which business rules apply.</li><li>What standards govern each task.</li></ul><p>Without that context, agents may produce technically plausible work that does <strong>not align with the company’s architecture or business needs.</strong> This is why <strong>agentic software delivery</strong> needs a <strong>connected</strong> <strong>knowledge layer.</strong></p><p>In many organizations, project knowledge is scattered across documentation, repositories, tickets, business systems, and the people who have worked on the applications for years. When a task moves from one team to another, someone often has to explain the same requirements, decisions, and constraints all over again.</p><p>A <strong>knowledge layer</strong> brings that information together and makes it accessible to the workflows where agents operate.</p><p>Basically, it can connect requirements, architecture decisions, code changes, incidents, business rules, and historical project context. <strong>Knowledge graphs</strong> and <strong>connected repositories</strong> can help establish those relationships and make the information easier to retrieve and use.</p><p>Over time, this becomes a form of <strong>organizational memory.</strong> Agents can draw on existing context instead of starting from scratch with every task, <strong>reducing the need for repeated explanations</strong> and <strong>preserving knowledge</strong> that might otherwise remain scattered across teams.</p><p>Finally, the <a href="https://inclusioncloud.com/insights/blog/what-is-the-ai-harness-model/"><strong>AI harness</strong></a><strong> </strong>connects this knowledge to the <strong>tools and workflows agents use</strong> to get work done. With it, we can provide the surrounding infrastructure that gives agents access to repositories, development environments, testing tools, and relevant enterprise data, while defining what they can access, which actions they can take, and when human review is required.</p><p>Together, the knowledge layer and the harness help agents work with the context, tools, and guardrails they need. This makes it easier to move from isolated AI interactions to repeatable workflows, while allowing agents to handle more structured tasks independently and escalate decisions that require human input.</p><h3>AI Pods: Compact Teams Built for Continuous Software Delivery</h3><p>As AI agents take on more structured engineering tasks, <strong>software delivery teams</strong> can become smaller and more focused.</p><p><strong>Traditional development </strong>often relies on larger teams with specialized roles coordinating work across multiple stages. An <strong>agentic approach</strong> changes how that capacity is organized. A compact group of experienced engineers can supervise specialized agents that handle parts of the execution process in parallel or across different stages.</p><p>This is the approach we took for our <a href="https://inclusioncloud.com/services/ai-pods/"><strong>Inclusion Cloud AI Pods.</strong></a></p><p>We bring <strong>senior engineering professionals</strong> and <strong>specialized AI agents</strong> together in <strong>compact delivery teams</strong> supported by reusable workflows and coordinated execution. The goal is to combine the technical judgment and <strong>domain expertise of experienced engineers</strong> with the <strong>execution capacity of AI agents.</strong></p><p>Within an AI Pod, <strong>agents can take on structured activities</strong> such as code generation, documentation, testing, data processing, and workflow execution. <strong>Engineers establish technical direction</strong>, provide business and architectural context, coordinate agent activities, and review their outputs.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/819/0*owgAc19MNmNlw7Ku.png" /></figure><p>This allows a smaller team to handle a broader range of tasks without adding a separate person for every activity. And <strong>time-zone alignment</strong> can extend that advantage even further.</p><p>When our <strong>engineering teams</strong> work in <strong>compatible time zones </strong>with their clients, they can coordinate priorities, review progress, anticipate upcoming discussions, and resolve questions throughout the working day. <strong>AI agents</strong> can then continue <strong>executing assigned tasks beyond those hours</strong>, preparing for the next review cycle.</p><p>This creates a delivery model where work can progress across a <strong>24-hour cycle.</strong> People provide direction and oversight during their working hours, while agents continue with structured tasks outside those hours.</p><h3>Enterprise AI Accelerators: Build on What Already Works</h3><p>AI Pods provide the people and execution capacity behind agentic software delivery. But teams also need a practical way to <strong>build and deploy solutions without starting from scratch</strong> every time.</p><p><strong>Enterprise AI initiatives</strong> often require connecting applications and data sources, adapting workflows to business requirements, configuring permissions, validating outputs, and meeting operational standards. These steps can take significant time, especially when each implementation begins with a blank slate.</p><p><strong>Reusable AI assets</strong> and<strong> implementation frameworks</strong> can help reduce that effort.</p><p>This is the principle behind our <a href="https://inclusioncloud.com/services/ai-accelerators/"><strong>Enterprise AI Accelerators.</strong></a><strong> </strong>They provide reusable technology components, integration patterns, and solution foundations that can be configured for an organization’s systems, data, and business processes.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/819/0*c3V6ydHZ9lMJdGUr.png" /></figure><p>Rather than rebuilding common capabilities for every initiative, teams can start with established foundations and focus their effort on what makes each implementation different. Our approach follows <strong>three stages:</strong></p><h4>1. DEFINE: Identify the right use case and implementation path.</h4><p>The process starts by understanding the business problem, existing systems, data requirements, and expected outcomes. This includes assessing the enterprise architecture, identifying integration points, defining governance requirements, and determining which reusable assets can support the initiative.</p><h4>2. BUILD: Configure and integrate the solution.</h4><p>Reusable components and implementation patterns provide a starting point for development. They are then adapted to the organization’s applications, data structures, business rules, and workflows. Integration, testing, and validation ensure that the solution works within the broader enterprise environment and meets the required technical and operational standards.</p><h4>3. SCALE: Move from implementation to production.</h4><p>Once validated, the solution can be deployed into production, monitored, and refined as business requirements evolve. Reusable foundations also make it easier to extend successful implementations to additional processes, applications, or business units.</p><p>Combined with AI Pods, they add a complementary <strong>technology layer</strong> to <strong>agentic software delivery.</strong></p><p>The first ones <strong>organize the people, agents, and workflows</strong> responsible for execution. Meanwhile, Accelerators provide <strong>reusable implementation foundations</strong> that help those teams configure, integrate, and deploy solutions more efficiently.</p><h3>Bringing Together an AI-Native Software Delivery Model</h3><p>At the end, building an effective <strong>AI-native software delivery model</strong> requires bringing together the organizational and technological capabilities behind agentic software delivery.</p><p><strong>AI Pods</strong> and <strong>AI Accelerators</strong> form complementary components of our <strong>AI-native software delivery model</strong>, combining agent-enabled execution with reusable enterprise technology foundations. Each addresses a different <strong>dimension of the delivery process:</strong></p><ul><li><strong>AI Pods define how work gets done</strong>, bringing together senior engineering professionals, specialized AI agents, domain expertise, and reusable workflows to coordinate execution and maintain human oversight.</li><li><strong>AI Accelerators define what makes implementation repeatable</strong>, providing reusable components, integration patterns, and solution foundations to configure, validate, and deploy AI capabilities within existing enterprise environments.</li></ul><p>Together, they connect <strong>execution capacit</strong>y with <strong>implementation foundations.</strong> AI Pods can lead the development and integration of enterprise solutions, while accelerators provide proven assets that reduce repetitive work and support consistent implementation.</p><p>This combination can <strong>support the entire journey of agentic software delivery</strong>, from identifying an initial use case and building a working solution to integrating it with enterprise systems, deploying it into production, and extending it across additional processes and business units.</p><p>The result is an<strong> integrated approach</strong> to <strong>AI-native software delivery</strong>, where specialized professionals and AI agents work with reusable technology foundations to turn enterprise AI initiatives into scalable capabilities.</p><p>If your organization is exploring how to evolve its software delivery model with Agentic AI, <a href="https://inclusioncloud.com/contact-us/"><strong>schedule a discovery call</strong></a> with our team to discuss your priorities and potential implementation paths.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=569f2a708319" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[SAP WM End of Support 2027: Deadlines & Migration Guide]]></title>
            <link>https://medium.com/@inclusion.cloud/sap-wm-end-of-support-2027-deadlines-migration-guide-e6786ea2aef4?source=rss-aa686e488265------2</link>
            <guid isPermaLink="false">https://medium.com/p/e6786ea2aef4</guid>
            <category><![CDATA[sap]]></category>
            <category><![CDATA[warehouse-management]]></category>
            <category><![CDATA[sap-migration]]></category>
            <category><![CDATA[data-migration]]></category>
            <category><![CDATA[software-engineering]]></category>
            <dc:creator><![CDATA[Inclusion Cloud ]]></dc:creator>
            <pubDate>Fri, 30 Oct 2026 18:33:02 GMT</pubDate>
            <atom:updated>2026-10-30T18:33:02.012Z</atom:updated>
            <content:encoded><![CDATA[<p>For years, SAP Warehouse Management (WM) has helped manufacturers, distributors, and logistics companies manage inventory, warehouse movements, and daily operations.</p><p>But SAP’s warehouse management portfolio is changing. With the compatibility scope deadline already behind us and mainstream maintenance for SAP Business Suite 7 ending in 2027, companies running older SAP warehouse solutions have some important decisions ahead.</p><p><strong>Does SAP WM still have a future? Is moving to SAP Extended Warehouse Management (EWM) mandatory? And what migration options are available?</strong></p><p>Let’s review the current status of SAP Warehouse Management, the deadlines that matter, and how companies can prepare for the next phase.</p><figure><img alt="SAP warehouse management migration" src="https://cdn-images-1.medium.com/max/1024/1*CPuaNxxHRYnVxP51ye1SJw.png" /></figure><h3>How SAP Warehouse Management Has Evolved</h3><p>SAP’s warehouse management capabilities date back to the SAP R/3 era, when Warehouse Management became part of the Logistics Execution (LE) functionality.</p><p>Known as <strong>SAP LE-WM</strong>, the solution allowed companies to manage storage bins, goods movements, stock transfers, picking, and physical inventory within their ERP environment.</p><p>As warehouse operations grew more complex, SAP introduced <a href="https://www.sap.com/products/scm/extended-warehouse-management.html"><strong>Extended Warehouse Management (EWM)</strong></a> in the mid-2000s, expanding support for advanced picking strategies, resource management, cross-docking, yard management, and warehouse automation.</p><p>The portfolio evolved again with SAP S/4HANA, introducing new deployment options and a clearer distinction between basic warehouse operations and more sophisticated logistics environments.</p><p>Today, three solutions are particularly relevant:</p><ul><li><strong>SAP LE-WM:</strong> The traditional warehouse management functionality used in SAP ECC and, temporarily, within the S/4HANA compatibility scope.</li><li><strong>SAP Stock Room Management:</strong> Introduced with S/4HANA 1909 to preserve essential LE-WM capabilities for simpler warehouse operations.</li><li><strong>SAP Extended Warehouse Management (EWM):</strong> SAP’s strategic warehouse management solution, available in embedded and decentralized deployment models.</li></ul><p>Understanding where each solution stands is the first step toward planning a migration.</p><h3>SAP Warehouse Management Deadlines: What Companies Need to Know</h3><p><a href="https://inclusioncloud.com/insights/blog/sap-migration-integration-suite/">Several deadlines</a> affect SAP warehouse management systems, although they don’t all apply to the same products or customers.</p><figure><img alt="SAP Warehouse Management versions deadline" src="https://cdn-images-1.medium.com/max/1024/1*79aFD8qdVTl-dFAo3SQViQ.png" /></figure><h3>2025–2026: The SAP LE-WM Compatibility Scope Deadline</h3><p>When SAP introduced S/4HANA, it allowed customers to continue using certain classic ERP functionalities through temporary compatibility rights.</p><p>This included advanced LE-WM capabilities that weren’t part of SAP’s permanent S/4HANA scope.</p><p>These rights were originally scheduled to expire on December 31, 2025. SAP subsequently announced a final five-month transition period for eligible on-premise Compatibility Packs, extending their usage rights through <strong>May 31, 2026</strong>.</p><p>That deadline has now passed.</p><p>Companies that migrated to S/4HANA while retaining advanced LE-WM functionalities should review which components remain in use and whether their contractual arrangements permit continued operation.</p><p>Certain <a href="https://inclusioncloud.com/insights/blog/rise-with-sap-partners-role/">RISE with SAP</a> agreements may have different transition provisions.</p><p>Importantly, this doesn’t mean that all classic warehouse management functionality has disappeared from S/4HANA. Stock Room Management continues to support eligible basic warehouse processes.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FWWWYeaiW9b8%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DWWWYeaiW9b8&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FWWWYeaiW9b8%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/7f55b37e1c586939fed37f9f3cd75a8d/href">https://medium.com/media/7f55b37e1c586939fed37f9f3cd75a8d/href</a></iframe><h3>2027: The SAP ECC Maintenance Deadline</h3><p>The next major deadline is <strong>December 31, 2027</strong>, when mainstream maintenance ends for eligible SAP Business Suite 7 applications.</p><p>This matters for companies still using SAP LE-WM within ECC, since the maintenance deadline affects the ERP environment supporting their warehouse operations.</p><p>It also affects customers running SAP SCM EWM 9.5, an older EWM version based on the pre-S/4HANA architecture.</p><p>For both groups, the approaching deadline creates a reason to evaluate their warehouse management roadmap alongside the broader modernization strategy.</p><h3>2030: Extended Maintenance Comes to an End</h3><p>SAP offers optional extended maintenance for eligible Business Suite 7 applications through December 31, 2030.</p><p>This provides additional time for companies that aren’t ready to complete their migration by 2027, although extended maintenance comes at an additional cost.</p><p>After that date, affected systems move to customer-specific maintenance, which offers a more limited level of support.</p><p>For companies operating older SAP warehouse solutions, these milestones make the coming years an important planning period.</p><h3>Does SAP Stock Room Management Have an End-of-Life Date?</h3><p>One common misunderstanding is that every company using classic SAP WM must immediately migrate to EWM.</p><p>That’s not necessarily the case.</p><p>SAP introduced <strong>Stock Room Management</strong> to preserve essential warehouse functionality within S/4HANA beyond the compatibility scope deadline.</p><p>It supports processes such as storage bin management, goods receipts and issues, internal stock movements, and physical inventory.</p><p>Unlike the advanced LE-WM components excluded from its scope, Stock Room Management doesn’t have a specific announced retirement deadline.</p><p>However, SAP has made its development strategy clear: no significant new functional development is planned for Stock Room Management.</p><p>Advanced capabilities such as Task and Resource Management, Yard Management, Cross-Docking, Wave Management, and certain warehouse automation interfaces are outside its scope.</p><p>For companies relying on those processes, SAP EWM is the strategic alternative.</p><h3>What Migration Paths Are Available for SAP Customers?</h3><p>The right approach depends on the current SAP landscape, warehouse complexity, and long-term business needs.</p><p>There are three main scenarios to consider.</p><h4>Path 1: From SAP LE-WM to Stock Room Management</h4><p>For companies running SAP ECC with relatively simple warehouse operations, moving to Stock Room Management within S/4HANA may be the most practical approach.</p><p>Stock Room Management preserves much of the familiar LE-WM functionality, allowing organizations to continue managing basic warehouse processes without adopting the full EWM feature set.</p><p>SAP notes that transitioning from LE-WM to Stock Room Management doesn’t require a separate technical warehouse management migration.</p><p>However, the broader ECC-to-S/4HANA transition still needs to be planned and executed, and companies must confirm that their existing processes fit within the supported Stock Room Management scope.</p><p><strong>Best suited for:</strong> Companies with straightforward warehouse operations that don’t require advanced automation, labor management, or sophisticated logistics processes.</p><h4>Path 2: From SAP LE-WM to SAP EWM</h4><p>Companies running more complex warehouse operations may choose to migrate directly from classic LE-WM to SAP EWM.</p><p>This allows them to adopt more advanced capabilities, including resource management, wave planning, warehouse automation, and additional options for integrating logistics processes.</p><p>Within S/4HANA, EWM can be deployed in two main ways:</p><ul><li><strong>Embedded EWM:</strong> Runs within the S/4HANA system, bringing warehouse and ERP processes together.</li><li><strong>Decentralized EWM:</strong> Runs on a separate system, providing architectural independence for complex warehouse environments or multi-ERP landscapes.</li></ul><p>Unlike Stock Room Management, moving from LE-WM to EWM involves a change in application architecture and warehouse process configuration.</p><p><strong>Best suited for:</strong> Companies with complex logistics requirements, automated warehouses, high transaction volumes, or plans to expand warehouse capabilities.</p><h4>Path 3: From Legacy SAP EWM to EWM on S/4HANA</h4><p>Some organizations already use SAP Extended Warehouse Management, but on older SAP SCM infrastructure.</p><p>For these customers, the migration involves moving to EWM based on S/4HANA rather than replacing classic LE-WM.</p><p>This can preserve existing investments in EWM processes while modernizing the underlying architecture and preparing the environment for ongoing SAP innovation.</p><p>SAP provides migration tools for supported EWM master data and system settings, although technical dependencies and customer-specific developments must still be reviewed.</p><p><strong>Best suited for:</strong> Companies using SAP SCM EWM 9.x that need to modernize their platform and align with SAP’s long-term maintenance strategy.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8L7zb5e8SGqu1M_kQq3Saw.png" /></figure><h3>How SAP Migration Cockpit Can Simplify the Transition</h3><p>SAP provides migration tools designed to reduce the effort involved in transferring warehouse data to EWM.</p><p>One of the most relevant is the <a href="https://www.sap.com/documents/sap-s4hana-migration-cockpit-migrate-your-data-to-sap-s4hana-1078635">SAP S/4HANA Migration Cockpit</a>, which supports structured data migration using predefined migration objects.</p><p>For LE-WM to EWM migrations, SAP provides capabilities for transferring warehouse information such as storage bins, stock, fixed-bin assignments, and other supported warehouse data.</p><p>Depending on the source and target releases, the migration process can include:</p><ol><li>Extracting warehouse data from the source system.</li><li>Loading the data into staging tables.</li><li>Mapping source fields to the target EWM structure.</li><li>Simulating the migration to identify potential errors.</li><li>Executing the migration and validating the results.</li></ol><p>SAP also provides <strong>direct-transfer migration objects</strong> for supported scenarios involving older EWM systems.</p><p>These capabilities can reduce repetitive manual work, improve consistency, and help teams identify data issues before the production cutover.</p><p>For example, SAP documents a process that uses transaction /SCWM/MIG_FILE_MC and the Migration Cockpit to transfer storage bin information from LE-WM to EWM.</p><p>However, the available migration objects and supported approaches depend on the SAP release and deployment model.</p><p>The Migration Cockpit is an important accelerator, but its scope is primarily focused on supported data and settings rather than the complete transformation of warehouse processes.</p><h3>Why a Migration Still Requires Specialized Expertise</h3><p>Even with SAP’s migration tools, moving from LE-WM to EWM can involve significant <strong>functional and technical work</strong>.</p><p>SAP’s migration guidance explicitly identifies several areas that aren’t automatically transferred, including existing RF transactions, user authorization roles, customer-specific developments, and much of the application configuration.</p><p>Transactional data and open warehouse documents also require careful handling, since the standard migration tools don’t simply move every active warehouse transaction to the new environment.</p><p>This creates several workstreams that need specialized attention.</p><ol><li><strong>Warehouse process design.</strong> EWM has a different architecture and process model from LE-WM. Existing picking, putaway, replenishment, and inventory processes may need to be redesigned rather than copied directly.</li><li><strong>Data preparation and validation.</strong> Warehouse master data, stock balances, storage bins, and related information need to be mapped, cleansed, migrated, and reconciled.</li><li><strong>Custom developments and integrations.</strong> Companies may have custom ABAP developments, RF applications, barcode scanners, conveyor systems, manufacturing integrations, and third-party logistics applications that require adaptation.</li><li><strong>Testing and cutover planning.</strong> Warehouse operations are time-sensitive. A migration must account for inventory accuracy, open transactions, operational continuity, performance testing, and the transition to the production environment.</li></ol><p>This is where working with an experienced SAP implementation partner can make a difference.</p><p>A partner can help organizations evaluate their existing landscape, design the target architecture, coordinate technical and functional workstreams, and manage the transition from planning through go-live and post-migration support.</p><h3>How Inclusion Cloud Can Support Your SAP WM Migration</h3><p>At Inclusion Cloud, we’ve been an <a href="https://inclusioncloud.com/technologies/sap-consulting-and-solutions/">SAP Partner for more than 20 years</a>, helping organizations implement, integrate, modernize, and support their enterprise environments.</p><p>Whether you’re planning a transition from classic SAP WM, modernizing an existing EWM landscape, or preparing for an S/4HANA migration, our teams can support different stages of the project.</p><p>Our capabilities include:</p><ul><li><strong>SAP assessments and migration planning:</strong> Reviewing the existing landscape, identifying dependencies, and defining the target architecture and migration roadmap.</li><li><strong>Implementation and integration:</strong> Supporting S/4HANA and warehouse-related projects, custom development, data migration, and connections with enterprise systems.</li><li><strong>Testing and deployment support:</strong> Assisting with validation, cutover preparation, issue resolution, and post-go-live stabilization.</li><li><strong>Certified SAP resources:</strong> Providing certified consultants and engineering resources through onshore, nearshore, and hybrid delivery models.</li></ul><h3>AI Pods: Compact Teams for SAP Modernization</h3><p>For projects that need specialized expertise and faster execution, Inclusion Cloud also offers <a href="https://inclusioncloud.com/services/ai-pods/"><strong>AI Pods</strong></a>.</p><p>AI Pods combine senior specialists with AI agents that can handle repeatable tasks, keep work moving beyond traditional working hours, and reduce unnecessary handoffs. Engineers remain involved throughout, setting direction, reviewing results, and ensuring quality.</p><p>In a warehouse modernization project, an AI Pod can support activities such as technical documentation, custom code analysis, data preparation, integration development, testing, and issue resolution.</p><p>This model allows organizations to add specialized delivery capacity without assembling a large team for every technical workstream.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*tf1yQLsKURI02hkTvJVq5Q.png" /></figure><h3>Enterprise AI Accelerators for SAP Migration</h3><p>Another way to simplify enterprise modernization is to reuse proven tools, components, and workflows rather than developing every capability from scratch.</p><p>That’s the idea behind Inclusion Cloud’s <a href="https://inclusioncloud.com/services/ai-accelerators/"><strong>Enterprise AI Accelerators</strong></a>: configurable assets combined with specialized engineering expertise to speed up implementation.</p><p>For SAP migration projects, a migration-focused accelerator can be configured around activities such as:</p><p>These capabilities complement SAP’s own tools, including the Migration Cockpit, and help streamline the work that falls outside standard automated migration. And the best part? Companies can <strong>start a few steps ahead</strong> with prebuilt assets and AI-enabled workflows, moving faster without the costs and overhead of assembling a large migration team.</p><h3>Planning Your Next Move</h3><p>Planning an SAP WM migration or considering a move to EWM? Whether you’re just exploring your options or already have a project underway, we can help you find the right approach and move forward with the right expertise and resources.</p><p><a href="https://inclusioncloud.com/contact-us/"><strong>Let’s discuss your SAP modernization roadmap.</strong></a></p><p><em>Originally published at </em><a href="https://inclusioncloud.com/insights/blog/sap-warehouse-management-migration/"><em>https://inclusioncloud.com</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e6786ea2aef4" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[What’s New in AI Security and Governance at Oracle AI World 2026?]]></title>
            <link>https://medium.com/@inclusion.cloud/whats-new-in-ai-security-and-governance-at-oracle-ai-world-2026-1afd5adc87e8?source=rss-aa686e488265------2</link>
            <guid isPermaLink="false">https://medium.com/p/1afd5adc87e8</guid>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[oracle]]></category>
            <category><![CDATA[openai]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[agentic-ai]]></category>
            <dc:creator><![CDATA[Inclusion Cloud ]]></dc:creator>
            <pubDate>Wed, 28 Oct 2026 14:10:21 GMT</pubDate>
            <atom:updated>2026-10-28T14:10:21.528Z</atom:updated>
            <content:encoded><![CDATA[<p><em>“Now that AI works, what are we going to do with it?”</em></p><p>That was the question behind Oracle CEO Clay Magouyrk’s keynote on Day 2 of Oracle AI World 2026 in Las Vegas.</p><p>The previous day’s announcements, including Oracle Fusion Claw, focused on what AI agents can accomplish inside enterprise applications: investigating problems, optimizing operations, and executing complex workflows.</p><p>But giving AI agents more autonomy introduces new questions. How do you control what they can access? Who approves their actions? How do you find vulnerabilities before attackers exploit them? And how do agents understand the business context behind the data they use?</p><p>On October 27, Oracle introduced several capabilities addressing these challenges, including <strong>Oracle AI Security Mesh, new identity and access controls, and major updates to Oracle AI Data Platform</strong>.</p><p>Here’s a closer look at the announcements.</p><figure><img alt="Oracle AI World 2026 Sam Altman" src="https://cdn-images-1.medium.com/max/1024/1*LCFsMSD6nX_lgchPKRHTNw.jpeg" /></figure><h3><strong>1. Oracle AI Security Mesh: Using AI to find and fix vulnerabilities</strong></h3><p>One of the biggest announcements was <strong>Oracle AI Security Mesh</strong>, a new managed security platform powered by OpenAI’s advanced cybersecurity models.</p><p>During the keynote, Clay Magouyrk described how its internal security practices have evolved as AI models have become more capable.</p><p>The company reported that it now discovers and remediates vulnerabilities in software and service configurations <strong>10 times faster than six months earlier</strong>.</p><p>That improvement matters because AI is changing the cybersecurity landscape for both attackers and defenders. Models can analyze code, identify potential weaknesses, and investigate how vulnerabilities might be exploited.</p><p>Oracle AI Security Mesh aims to make these capabilities available to enterprise security teams.</p><h4><strong>How Oracle AI Security Mesh works</strong></h4><p>The platform is designed to support a continuous security workflow:</p><ol><li><strong>Discover:</strong> Map applications, infrastructure, and potentially exposed systems across cloud and on-premises environments.</li><li><strong>Investigate:</strong> Use specialized AI models to analyze code, configurations, and possible attack paths.</li><li><strong>Validate:</strong> Test suspected weaknesses to determine whether they represent exploitable vulnerabilities.</li><li><strong>Remediate:</strong> Generate proposed fixes and validate their effects on application behavior.</li><li><strong>Deploy and verify:</strong> Apply approved changes and confirm that the vulnerability has been addressed.</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zkhRGfHqMPixlG2pIvPbOw.jpeg" /></figure><p>During the keynote, Oracle demonstrated the platform using a hypothetical manufacturing company.</p><p>The system discovered an issue in an ordering API that could allow unauthorized access to sensitive information through another misconfigured service.</p><p>Specialized agents investigated the attack path, validated the vulnerability, and prepared a remediation plan. A coding agent then generated the necessary corrections.</p><p>An important detail is that <strong>customers approve remediation changes before AI Security Mesh applies them</strong>. Agents also operate with distinct identities, limited task scopes, and temporary permissions.</p><p>For security teams, the potential value is a shorter path from discovering a vulnerability to deploying a verified fix, with human approval and technical controls built into the process.</p><h3><strong>2. Agent identity and access: Who is actually taking action?</strong></h3><p>Another major topic was how identity management needs to evolve when AI agents begin acting on behalf of employees and business systems.</p><p>Traditionally, identity and access management focuses on human users, applications, and service accounts.</p><p>AI agents introduce additional complexity. They may need to access several systems, request permissions, perform sensitive operations, and coordinate with other agents.</p><p>Giving every agent broad access through an employee’s account creates obvious problems. It becomes difficult to determine who performed an action, which permissions were necessary, and whether the activity was authorized.</p><p>Oracle addressed this challenge through several identity and access capabilities.</p><h4><strong>Dedicated identities for AI agents</strong></h4><p>Oracle discussed agent identities within its Identity and Access Management approach, allowing organizations to distinguish agent activity from human activity.</p><p>The objective is to support large numbers of agents without relying on shared user credentials.</p><p>Each agent can operate under its own identity, making it possible to associate permissions and actions with a particular agent.</p><h4><strong>Just-in-time permissions</strong></h4><p>Agents also need appropriate limits on their access.</p><p>Rather than granting broad, permanent permissions, Oracle emphasized access that can be:</p><ul><li>Limited to a specific task or resource.</li><li>Granted when needed.</li><li>Restricted to a short period.</li><li>Subject to additional approval for sensitive actions.</li></ul><p>This reduces the risks associated with agents having persistent access to enterprise systems.</p><h4><strong>Biometric verification for human approvals</strong></h4><p>Oracle also introduced biometric capabilities through OCI Identity Assurance.</p><p>These can include facial recognition, voice recognition, and fingerprint verification to confirm the identity of users approving sensitive actions.</p><p>The rationale is straightforward: if an AI agent requires human approval before executing an operation, the organization needs confidence that the approval comes from an authorized human.</p><p>This becomes particularly relevant for activities such as releasing payments or changing access permissions.</p><p>Together, dedicated agent identities, temporary permissions, and verified human approvals establish stronger boundaries for agentic workflows.</p><h3><strong>3. Oracle AI Data Platform: Giving agents the right business context</strong></h3><p>Security controls determine what an AI agent is allowed to do. But another question remains: <strong>does the agent understand enough about the business to make an appropriate decision?</strong></p><p>Enterprise data is distributed across ERP systems, databases, analytics platforms, documents, APIs, and external applications.</p><p>Simply making that information accessible to an AI model does not guarantee that it understands relationships, policies, or business definitions.</p><p>For example, an agent investigating an overdue invoice needs more than an invoice amount and due date.</p><p>It may also need to understand payment terms, customer relationships, approval rules, disputed transactions, and the relevant financial processes.</p><p>During the keynote, Oracle announced new <strong>AI-generated business ontologies</strong> within Oracle AI Data Platform.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*hpwzJ4IHlUvgLUDmzSeMWA.jpeg" /></figure><h4><strong>What are business ontologies?</strong></h4><p>Business ontologies define the concepts, entities, relationships, and rules that give enterprise information its meaning.</p><p>They help agents understand how different pieces of business information relate to each other and how those relationships affect decisions.</p><p>Oracle’s AI Data Platform uses an AI Discovery agent to analyze information such as database schemas, application documentation, semantic models, APIs, and enterprise documents.</p><p>It can then propose business ontologies that domain experts review, refine, and approve.</p><p>This human-governed approach is important because the business remains responsible for defining what its information means.</p><h4><strong>Semantic reasoning and reusable business knowledge</strong></h4><p>Oracle also introduced capabilities built around its Semantic Reasoning Engine.</p><p>The engine uses enterprise ontologies to assemble relevant business context before information reaches the AI model.</p><p>This helps agents interpret business relationships, retrieve information across systems, and apply consistent definitions.</p><p>Reusing this context may also reduce repetitive reasoning and unnecessary token consumption.</p><p>For enterprises deploying multiple AI agents, the broader opportunity is to build a shared layer of business knowledge rather than requiring every agent to interpret the organization’s data independently.</p><h3><strong>4. Governing an expanding enterprise AI environment</strong></h3><p>As organizations deploy more agents, governance becomes increasingly difficult.</p><p>Different departments may create agents for finance, HR, procurement, operations, and customer service. Each agent may use different tools, data sources, permissions, and models.</p><p>Without a centralized inventory, companies can lose visibility into what agents exist, who owns them, and what they are allowed to do.</p><p>Oracle announced several AI Data Platform governance capabilities to address this challenge.</p><h4><strong>Oracle AI Registry</strong></h4><p>Oracle AI Registry is intended to maintain a centralized inventory of enterprise AI components, including:</p><ul><li>AI agents and their owners.</li><li>Tools, skills, and MCP servers.</li><li>Versions and dependencies.</li><li>Permissions and lifecycle status.</li></ul><p>This provides a foundation for understanding which AI capabilities are deployed and how they are connected.</p><h4><strong>Oracle AI Governance and Control</strong></h4><p>Oracle also outlined planned governance capabilities for testing, policy assessment, and production monitoring.</p><p>These are designed to help organizations evaluate agent behavior, quality, performance, usage, and costs.</p><p>That includes identifying unexpected behavior and supporting corrective actions as AI systems evolve.</p><p>The distinction between announced capabilities and their availability matters here: Oracle said the new AI Data Platform capabilities are expected to become available over the next 12 months.</p><p>For enterprise teams, the announcement points toward more centralized oversight of agents, their dependencies, and their operating costs.</p><h3><strong>5. Sam Altman on AI security: Why companies need to act now</strong></h3><p>Another highlight of the keynote was the conversation between Clay Magouyrk and OpenAI CEO Sam Altman.</p><p>The discussion covered AI-assisted software development, the accelerating capabilities of AI models, and the implications for cybersecurity.</p><p>Altman emphasized the importance of understanding the changing AI security threat landscape and addressing emerging risks.</p><p>One point from the discussion is particularly relevant: AI is becoming capable of discovering vulnerabilities that may have existed in software for years.</p><p>This creates an advantage for organizations that use those same capabilities defensively.</p><p>The challenge is that vulnerabilities can be discovered at a pace that traditional manual investigation and remediation processes may struggle to match.</p><p>That makes automated security testing, validation, and remediation increasingly important.</p><p>It also helps explain Oracle’s approach with AI Security Mesh: using advanced AI models to strengthen the systems that other AI agents and business applications depend on.</p><h4><strong>What these announcements mean for Oracle customers</strong></h4><p>Together, the Day 2 announcements address several practical requirements for enterprise AI adoption.</p><p><strong>Security:</strong> Organizations need ways to identify and remediate vulnerabilities across the systems their agents interact with.</p><p><strong>Identity and permissions:</strong> Agents need identifiable ownership, limited access, and appropriate human approvals.</p><p><strong>Business context:</strong> Agents need to interpret enterprise information according to the relationships, definitions, and processes that govern the business.</p><p><strong>Operational governance:</strong> IT teams need visibility into deployed agents, their dependencies, behavior, and costs.</p><p>These capabilities are especially relevant for companies expanding their use of Oracle Fusion Applications, OCI, Oracle databases, and third-party enterprise platforms.</p><p>They also complement the capabilities introduced with Oracle Fusion Claw.</p><p>As agents take on more complex responsibilities, the supporting architecture becomes increasingly important. Data integration, identity management, security testing, business semantics, and monitoring all influence how reliably those systems can operate.</p><h3><strong>Bringing governed AI into your Oracle environment</strong></h3><p>At Inclusion Cloud, we’re an <a href="https://inclusioncloud.com/technologies/oracle-consulting-and-solutions/">Oracle Partner</a> with more than 20 years of experience supporting Oracle implementations, integrations, optimization, and specialized IT staffing.</p><p>Whether you’re evaluating new AI capabilities, connecting enterprise data, or expanding your Oracle delivery team, our specialists can help you plan and execute your next steps.</p><p><a href="https://inclusioncloud.com/technologies/oracle-consulting-and-solutions/"><strong>Book a discovery call with Inclusion Cloud</strong></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1afd5adc87e8" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[What Is Oracle Fusion Claw? A Look at Oracle’s New AI Capabilities]]></title>
            <link>https://medium.com/@inclusion.cloud/what-is-oracle-fusion-claw-a-look-at-oracles-new-ai-capabilities-8c2a47065ebf?source=rss-aa686e488265------2</link>
            <guid isPermaLink="false">https://medium.com/p/8c2a47065ebf</guid>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[oracle]]></category>
            <category><![CDATA[oracle-fusion-claw]]></category>
            <dc:creator><![CDATA[Inclusion Cloud ]]></dc:creator>
            <pubDate>Mon, 26 Oct 2026 19:54:22 GMT</pubDate>
            <atom:updated>2026-10-26T20:08:28.384Z</atom:updated>
            <content:encoded><![CDATA[<p>Our Inclusion Cloud team is in Las Vegas for <strong>Oracle AI World 2026</strong>, taking place at The Venetian Convention and Expo Center.</p><p>Day 1 brought several announcements for Oracle Fusion Applications, with Fusion Claw taking center stage.</p><p>During the opening keynote, Mike Sicilia and Chris Leone showed how Oracle is bringing AI deeper into everyday business operations. Later, Leone, Oracle’s EVP of Applications Development, went into more detail about how Fusion Claw works, with some interesting demos and use cases.</p><p>The idea behind Fusion Claw is pretty simple: instead of using AI just to help employees get things done, Oracle wants AI to take on more of the work itself. But there’s a challenge: how do you make that happen while keeping AI reliable, giving it the right business context, and making sure the costs don’t get out of hand?</p><p>That’s what Oracle is trying to address with <strong>Fusion Claw</strong>.</p><p>Let’s take a look at what was announced and how it works.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*k4yQrOJM-Cqk7N2FwqUVKQ.jpeg" /></figure><h3><strong>What Is Oracle Fusion Claw, and Why Is Oracle Introducing It?</strong></h3><p><a href="https://www.oracle.com/news/announcement/ai-world-oracle-introduces-claw-powered-fusion-agentic-applications-for-supply-chain-2026-10-26/">Oracle Fusion Claw</a> is what the company calls a <strong><em>governed agentic runtime</em></strong>. In simpler terms, it allows AI agents to perform complex work inside Oracle Fusion Applications, including tasks traditionally handled by financial analysts, supply chain planners, and other specialists.</p><p>Oracle has already introduced generative AI features, AI agents, and agentic applications across Fusion. But these capabilities have largely focused on assisting employees or coordinating specific workflows.</p><p>Fusion Claw<strong> extends</strong> that approach to more demanding work.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*a50FeO7btjxtkuutTmNGbw.jpeg" /></figure><blockquote><strong>“We do the work, we do the analysis, we do the deep research,”</strong> <em>Leone explained during his session.</em></blockquote><p>For example, Fusion Claw can analyze financial records, reconcile differences between ledgers, compare inventory scenarios, or optimize workforce schedules. It can examine millions of records and carry out processes that involve multiple steps.</p><p>Importantly, it doesn’t have <strong>unrestricted access</strong> to the system. It follows the company’s existing permissions, policies, and approval rules. When a decision falls outside those limits, it requires human review.</p><p>Leone also introduced the concept of <strong>“AI with receipts.”</strong> After completing a task, Fusion Claw produces an <em>outcome receipt</em>, showing what actions were taken, which rules were followed, and the results.</p><h3><strong>What Does Fusion Claw Look Like in Practice?</strong></h3><p>During the opening keynote, Chris Leone demonstrated how Fusion Claw could optimize transportation planning in Oracle Fusion SCM.</p><p>The system reviewed orders, shipping dates, truck capacity, and other constraints before proposing a new distribution plan.</p><p>In the demonstration, it increased projected truck utilization<strong> from 61% to 88%, with estimated savings of $58,000</strong>. Five loads required human review, while another 149 could be processed automatically.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ivN7AlVdPS3J2UojDR5UKw.jpeg" /></figure><p>A second demonstration focused on workforce scheduling in healthcare.</p><p>Fusion Claw evaluated staffing availability, employee qualifications, overtime costs, and scheduling restrictions. It compared three possible solutions and selected one that filled<strong> 70 of 72 open shifts, increasing coverage to 99% with a 5% increase in relative costs</strong>.</p><p>In both examples, the important distinction was that Fusion Claw could go beyond recommending a solution and execute the approved changes within Fusion.</p><h3><strong>Where Is Fusion Claw Available?</strong></h3><p>Oracle announced <strong>75 packaged agentic applications</strong>, including 25 already enabled with Fusion Claw, covering processes across Fusion ERP, SCM, HCM, and CX.</p><p>These applications are designed around specific business objectives, such as keeping financial operations on track, improving transportation planning, or maintaining adequate staffing levels.</p><p>Organizations can also create their own applications and Fusion Claw extensions through <a href="https://inclusioncloud.com/insights/blog/oracle-ai-agent-studio/">Oracle AI Agent Studio</a>, using either no-code tools or a development environment.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*52ELtQvyuSxgxz8RRqhkVQ.jpeg" /></figure><h3><strong>How Does Fusion Claw Work With Oracle’s Other Announcements?</strong></h3><p>Leone also introduced two other capabilities that help explain Oracle’s broader plans for Fusion.</p><p>The first is <strong>Fusion Companion</strong>, which changes how employees interact with the software.</p><p>Instead of navigating different applications, reports, and workflows, users can simply describe what they want to accomplish. Fusion Companion determines whether the request requires a standard transaction, an AI agent, or a more complex task handled by Fusion Claw.</p><p>It also works through Fusion Agentic Home, Oracle’s new interface for Fusion Applications, including mobile and voice interactions. Oracle demonstrated how users could delegate work without having to identify which application should handle it.</p><p>The second announcement was the <strong>Fusion Setup and Configuration Registry</strong>, an open-source collection of Oracle-maintained AI skills for setting up and configuring Fusion Applications.</p><p>The registry contains product knowledge, configuration requirements, dependencies, and recommended practices that AI coding agents can use during implementations.</p><p>As Leone explained, “Agents will be the operator, and the registry will be the authority.”</p><p>The goal is to reduce the manual work involved in implementing and configuring Fusion, while giving implementation teams more time to focus on business requirements and design decisions.</p><h3><strong>What Changes for Oracle Customers?</strong></h3><p>Fusion Claw<strong> adds another level of automation</strong> to Oracle Fusion Applications.</p><p>Employees can define what needs to be accomplished, review proposed actions when necessary, and allow the system to carry out work within established rules.</p><p>Alongside Fusion Companion and the Setup and Configuration Registry, Oracle is addressing two challenges at once: reducing the effort required to implement its applications and reducing the manual work needed to operate them.</p><p>As an <a href="https://inclusioncloud.com/technologies/oracle-consulting-and-solutions/">Oracle Partner</a> with 20+ years of experience, Inclusion Cloud helps companies implement, integrate, optimize, and support Oracle technologies, as well as scale their teams with certified resources.</p><p>Our team is at Oracle AI World 2026, so if you don’t want to miss the latest announcements and insights from the event, follow us on <a href="https://www.linkedin.com/company/2409140">LinkedIn</a>.</p><p>And if you’re looking for a partner to help bring trusted AI agents into your business, we’d be happy to discuss your project. <a href="https://inclusioncloud.com/contact-us/">Schedule a call with our team</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8c2a47065ebf" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[SAP BTP Services for US Enterprises in 2026]]></title>
            <link>https://medium.com/@inclusion.cloud/sap-btp-services-for-us-enterprises-in-2026-79db49bc961f?source=rss-aa686e488265------2</link>
            <guid isPermaLink="false">https://medium.com/p/79db49bc961f</guid>
            <category><![CDATA[sap]]></category>
            <category><![CDATA[sap-btp]]></category>
            <dc:creator><![CDATA[Inclusion Cloud ]]></dc:creator>
            <pubDate>Fri, 23 Oct 2026 15:30:05 GMT</pubDate>
            <atom:updated>2026-10-23T15:30:05.894Z</atom:updated>
            <content:encoded><![CDATA[<p>If your company runs on SAP, chances are you’ve already faced the challenge of connecting systems, adapting business processes, or adding new capabilities without making your ERP landscape harder to maintain.</p><p>That’s where <strong>SAP Business Technology Platform</strong> (BTP) comes in. It helps companies integrate applications, automate workflows, build custom solutions, and extend SAP without adding unnecessary complexity to the core.</p><p>And the business case is becoming harder to ignore. An<strong> </strong><a href="https://www.sap.com/documents/2025/06/86cf4594-0a7f-0010-bca6-c68f7e60039b.html"><strong>IDC study</strong></a> found that organizations using SAP BTP with SAP applications achieved a projected <strong>516% three-year ROI</strong>, with an average <strong>payback period of eight months.</strong></p><p>However, these are study-specific results, not a guarantee for every implementation. So where can BTP make the biggest difference for U.S. enterprises in 2026? The answer depends on what you’re trying to improve and how you approach implementation.</p><h3>What Are SAP BTP Services?</h3><p>Now, BTP is SAP’s platform for<strong> integrating systems</strong>, developing applications, <strong>automating processes</strong>, working with enterprise data, and <strong>building AI-powered business solutions.</strong> For enterprise IT teams, <strong>SAP BTP services</strong> generally fall into <strong>five areas:</strong></p><ul><li><strong>Integration:</strong> Connecting SAP and non-SAP applications, APIs, and business processes.</li><li><strong>Application development and extensions:</strong> Building custom applications and extending SAP functionality while supporting a clean-core strategy.</li><li><strong>Process automation:</strong> Automating approvals, repetitive tasks, and business workflows.</li><li><strong>Data and analytics:</strong> Making enterprise data more accessible and useful for reporting, planning, and AI.</li><li><strong>AI development and orchestration:</strong> Building AI-supported applications and agents that operate within enterprise workflows.</li></ul><p>These capabilities are connected, but they solve different problems. A company integrating SAP S/4HANA with Salesforce has different requirements from one automating invoice processing or developing an AI agent for procurement.</p><p>The important part is knowing which capabilities are needed, how they should work together, and what the company expects to achieve.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*g90PZ4VP4aljfIC7H7pRkw.png" /></figure><h3>What to Look for in an SAP BTP Services Partner</h3><p>A useful way to understand what an <strong>SAP BTP services partner</strong> should bring is to look at how these capabilities come together in practice.</p><p>At Inclusion Cloud, we combine <strong>20+ years of partnership</strong> experience with expertise in enterprise modernization, app development, system integration, process automation, and newer capabilities involving SAP Business Data Cloud and Joule. We are also an SAP <strong>Partner Center of Expertise (PCoE)</strong>, with established support infrastructure and processes.</p><p>But, beyond technical breadth, enterprise SAP projects require experience applying these capabilities in real business environments.</p><p>We have worked with Fortune 500 companies across multiple industries, and our broader delivery experience includes <strong>more than 200 clients</strong>, over <strong>5,000 projects</strong>, and <strong>300+ certified consultants</strong> across our technology practices.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FWWWYeaiW9b8%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DWWWYeaiW9b8&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FWWWYeaiW9b8%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/7f55b37e1c586939fed37f9f3cd75a8d/href">https://medium.com/media/7f55b37e1c586939fed37f9f3cd75a8d/href</a></iframe><p>In fact, our<strong> </strong><a href="https://inclusioncloud.com/technologies/sap-consulting-and-solutions/"><strong>SAP consulting and solutions practice</strong></a> provides an overview of these capabilities. On the other hand, our<strong> </strong><a href="https://inclusioncloud.com/clients-and-industries/case-studies/"><strong>client case studies</strong></a> illustrate how technology is applied to different business requirements.</p><p><strong>Talent</strong> is another consideration. <strong>SAP BTP projects</strong> may require different combinations of integration architects, application developers, automation specialists, data engineers, and AI expertise.</p><p>Through <a href="https://inclusioncloud.com/#:~:text=inMOVE%E2%84%A2%20by%20InclusionCloud"><strong>inMOVE™</strong></a>, our <strong>AI-powered talent identification</strong> and validation process, we combine AI-assisted screening with human review and technical interviews. The process draws on a database of <strong>more than one million profiles</strong> and can identify and present <strong>validated specialists in 72 hours</strong>, identifying the<strong> top 1% of senior SAP resources</strong> through technical screening and validation.</p><p>Taken together, these elements offer a <strong>practical framework</strong> for evaluating <strong>SAP BTP services partners:</strong></p><ul><li>Breadth of SAP expertise.</li><li>Experience delivering enterprise projects.</li><li>Established support processes and relevant case studies.</li><li>A credible approach to identifying and validating specialized talent.</li></ul><h3>What Works: The SAP BTP Services Worth Prioritizing in 2026</h3><p>There is no universal <strong>SAP BTP implementation roadmap.</strong> The right priorities depend on the company’s existing SAP landscape, the processes it wants to improve, and its broader technology strategy.</p><p>For many U.S. enterprises, however, <strong>five areas are particularly relevant.</strong></p><h4>1. SAP Integration Suite: Connecting Enterprise Systems</h4><p>First, we must notice that most companies operate a combination of SAP applications, third-party platforms, and legacy systems. The challenge is making those systems work together without creating a growing collection of fragile, difficult-to-maintain interfaces.</p><p><strong>SAP Integration Suite</strong> helps organizations connect applications and processes through API management, application integration, event-driven communication, and other integration capabilities. But let’s put it in a <strong>practical example.</strong></p><p>Consider a <strong>manufacturer</strong> running SAP S/4HANA alongside a separate warehouse management system, Salesforce, and a procurement platform. Information needs to move between these systems reliably, even when<strong> transaction volumes</strong> increase or <strong>individual applications</strong> change.</p><p>A well-designed <strong>integration architecture</strong> can standardize those connections, improve error handling, and make future integrations easier to maintain.</p><p>This is also becoming more important as companies <strong>introduce AI.</strong> An agent that needs to retrieve purchase order information, check inventory, or prepare a business transaction must interact with the relevant systems through reliable, appropriately secured interfaces.</p><p>That makes <strong>integration</strong> an architectural<strong> foundation for enterprise AI</strong>, not simply a technical task to complete before the AI project begins. So consulting may also include assessing legacy interfaces, rationalizing redundant connections, and planning migrations toward supported integration patterns.</p><h4>2. SAP BTP and Clean Core: Extending SAP S/4HANA Without Increasing Complexity</h4><p>Now, in second stance, one of the strongest reasons to consider SAP BTP services is the need to <strong>modernize ERP environments</strong> while <strong>preserving upgrade flexibility.</strong></p><p>Historically, companies often customized SAP systems directly to accommodate specific business requirements. Over time, those modifications could <strong>make upgrades more complicated</strong> and <strong>increase maintenance costs.</strong></p><p>The<strong> clean-core approach</strong> encourages organizations to reduce unnecessary modifications and use supported extension mechanisms wherever appropriate.</p><p>SAP BTP enables <strong>side-by-side extensions:</strong> applications and services that operate outside the ERP core and interact with it through supported interfaces. For example, a company might need a custom application for managing supplier exceptions.</p><p>Rather than embedding every requirement directly into SAP S/4HANA, it could build a separate application on BTP that connects to the ERP through APIs. This can help <strong>preserve the core system</strong> while allowing the business to develop functionality around its particular processes.</p><p>Of course, <strong>not every extension belongs on BTP.</strong> Some requirements can be handled through standard SAP functionality, while others may be better suited to ABAP Cloud extensions within the SAP environment.</p><p>The goal is not to move every customization outside the ERP. It is to choose an architecture that balances business flexibility, maintainability, and long-term upgrade requirements. For enterprises planning an <strong>SAP S/4HANA transformation</strong>, this is one of the most important decisions to make before development begins.</p><h4>3. SAP Build Process Automation: Moving Beyond Manual Work</h4><p>Many enterprise processes still depend on <strong>employees moving information manually</strong> between applications, checking documents, chasing approvals, and resolving exceptions manually. Accounts payable, procurement, employee onboarding, and operational reporting are common examples.</p><p><strong>SAP Build Process Automation</strong> helps organizations automate workflows, business rules, and repetitive tasks through a combination of workflow capabilities and robotic process automation. Let’s take invoice processing as an example.</p><p>A company may receive invoices through several channels, extract the relevant information, validate it against purchase orders, and route exceptions to finance employees.</p><p>A well-designed process can automate appropriate steps while keeping human review where information is incomplete, business rules require judgment, or an exception needs approval.</p><p>The <strong>value comes from the entire workflow</strong>, not simply from automating one task.</p><p>That means looking at processing time, error rates, manual effort, exception volumes, and the cost of maintaining the automation. But it also means understanding where <strong>other SAP capabilities</strong> (like <a href="https://inclusioncloud.com/insights/blog/sap-document-ai-workshop-recap/">SAP Document AI</a>) can <strong>complement workflow automation</strong> by extracting and validating information from incoming documents.</p><p>In short, if you’re exploring SAP BTP services, <strong>process automation</strong> is often a practical <strong>starting point</strong> because the business problem can be defined clearly and measured against an existing process.</p><h4>4. SAP BTP AI Services: From Experiments to Business Applications</h4><p><strong>Enterprise AI</strong> has introduced another set of <strong>requirements for SAP environments. </strong>However, we can summarize what companies want of AI applications in <strong>three main requirements:</strong></p><ol><li>Understand business context.</li><li>Access relevant enterprise information.</li><li>Help execute work rather than simply generate answers.</li></ol><p><strong>SAP’s AI capabilities</strong> support the <strong>development of applications</strong> and <strong>agents</strong> that can interact with enterprise data and business processes. Depending on the architecture and use case, this may involve SAP AI Core, Generative AI Hub, Joule Studio, SAP HANA Cloud, SAP Integration Suite, and SAP Build Process Automation.</p><p>In fact, SAP’s <a href="https://architecture.learning.sap.com/docs/golden-path/ai-golden-path/build-and-deliver/build-ai-agents"><strong>2026 architecture guidance for AI agents</strong></a> describes how these components can work together to build and connect AI agents. But what might this look like in practice? Let’s see three simple examples:</p><ul><li>A <strong>procurement workflow</strong> could use an agent to retrieve purchase order information, check a discrepancy against business rules, and prepare a proposed action for review.</li><li>A <strong>finance application</strong> could help employees investigate unmatched transactions by bringing together relevant payment, customer, and invoice information.</li><li>An<strong> operations solution</strong> could combine predictive insights with SAP business data to help prioritize maintenance activities.</li></ul><p>Now, these are examples of <strong>potential implementations</strong>, not turnkey capabilities that work without configuration. The <strong>difficult part</strong> is often connecting the AI to the right data, defining which actions it can perform, testing its behavior, and determining when a person must approve the result.</p><p>For that reason,<strong> SAP BTP AI projects</strong> need more than prompt engineering. They require application development, integration, enterprise architecture, security, testing, and an understanding of the underlying business process.</p><h4>5. SAP BTP Data and Analytics: Making Enterprise Data More Useful</h4><p>Finally, <strong>AI applications</strong> are only as useful as the <strong>information they can access</strong> and interpret.</p><p>Many enterprises have <strong>data distributed</strong> across ERP systems, data warehouses, cloud applications, and departmental platforms. Even when that data is technically available, <strong>inconsistent definitions</strong> and <strong>disconnected information</strong> can make it difficult to use reliably.</p><p><strong>SAP HANA Cloud</strong>, <strong>SAP Datasphere</strong>, and<strong> SAP Business Data Cloud </strong>can play different roles in a broader<strong> SAP data architecture</strong>, depending on the company’s requirements and existing investments. For example, a company may need to combine finance, customer, and operational information to understand profitability across business units.</p><p>An AI application may then use that information to answer questions, identify patterns, or support a workflow. But making the data available is only part of the problem.</p><p>The architecture must also account for data quality, permissions, business definitions, and the context needed to interpret results correctly. This is particularly important when AI agents are expected to take action based on enterprise information.</p><p>In short, when evaluating SAP BTP services,<strong> data architecture</strong> should be considered alongside<strong> integration</strong> and <strong>application development</strong> rather than treated as a separate concern.</p><h3>What Doesn’t Work: Implementing SAP BTP Without a Business Case</h3><p>Having access to a broad technology platform does not guarantee business value.</p><p>In fact, one of the easiest mistakes to make is to start with the available services rather than the problem that needs to be solved.</p><p>A company might introduce new integrations, develop custom applications, or launch an AI pilot without establishing what the initiative should improve.</p><p>The technology may work as designed, but the business case remains unclear.</p><p>There are three common reasons this happens.</p><p><strong>First, automation is added to a process that has not been redesigned.</strong></p><p>If an approval workflow contains unnecessary steps, inconsistent rules, and poorly defined responsibilities, automating it may simply reproduce those problems faster.</p><p><strong>Second, custom development introduces avoidable complexity.</strong></p><p>Not every requirement needs a new application. Standard SAP capabilities, existing APIs, prebuilt integration content, or simpler process changes may be enough.</p><p><strong>Third, AI is introduced before data and governance are ready.</strong></p><p>An agent cannot reliably make business decisions if it lacks the necessary context, has access to inconsistent information, or operates without clear boundaries.</p><p>The solution is not to avoid SAP BTP. It is to approach implementation with more discipline.</p><p>Before selecting a service, define the business outcome, map the current process, review the capabilities already available, and determine how success will be measured.</p><p>That approach also helps distinguish a useful proof of concept from a solution that is ready for production.</p><h3>How to Choose an SAP BTP Services Partner in the United States</h3><p>Selecting an <strong>SAP BTP consulting partner</strong> involves more than checking whether a provider has experience with the platform.</p><p>The right partner should be able to<strong> connect technical decisions</strong> with <strong>business requirements</strong> and understand how new capabilities will operate within the company’s existing SAP landscape. Here you have <strong>what is worth evaluating:</strong></p><h4><strong>1. Experience Across SAP and Non-SAP Systems</strong></h4><p>Integration work often becomes complicated at the boundaries between platforms.</p><p>Look for a partner with practical experience connecting <strong>SAP S/4HANA</strong> and <strong>other SAP applications</strong> with <strong>third-party</strong> enterprise systems, APIs, cloud services, and legacy environments.</p><p>Ask how the provider approaches error handling, monitoring, data consistency, and long-term maintenance, not just the initial connection.</p><h4><strong>2. Clean-Core and SAP S/4HANA Expertise</strong></h4><p>A good implementation partner should understand <strong>when to use SAP BTP side-by-side extensions</strong>, when ABAP Cloud is more appropriate, and when standard functionality is sufficient. The goal is to deliver the required business capability without creating unnecessary technical debt.</p><p>This is particularly important for companies preparing for SAP S/4HANA transformation or trying to simplify an existing ERP landscape.</p><h4><strong>3. Business Process Automation Experience</strong></h4><p><strong>Experience with SAP Build Process Automation</strong> should include process assessment, workflow design, business rules, exception handling, testing, and production monitoring.</p><p>A partner should be able to explain how the proposed automation will change the process and which metrics will demonstrate its value.</p><p>For <strong>document-heavy workflows</strong>, it may also be worth evaluating <strong>SAP Document AI</strong> and its role in extracting and validating information before downstream processing.</p><h4><strong>4. Practical Enterprise AI Capabilities</strong></h4><p>If <strong>AI is part of the roadmap</strong>, look for experience that extends beyond demonstrations.</p><p>The provider should understand enterprise data access, model selection, application integration, evaluation, security, human oversight, and production operations.</p><p>It should also be able to explain when an AI agent is appropriate and when a conventional workflow or deterministic automation would be more reliable.</p><h4><strong>5. A Flexible Delivery Model</strong></h4><p>Enterprise technology projects do not all require the same team structure.</p><p>Some initiatives need <a href="https://inclusioncloud.com/about/our-engineering-team/"><strong>senior architecture support</strong></a> and a small group of specialized engineers. Others require a broader implementation team or ongoing support across several systems.</p><p>A partner should be able to match the delivery model to the scope, maintain access to the right technical expertise, and provide clear ownership of the deliverables.</p><p>For U.S. companies, a combination of <a href="https://inclusioncloud.com/insights/blog/best-nearshore-sap-and-ai-firms/"><strong>onshore leadership</strong></a> and <strong>distributed engineering</strong> can also provide flexibility, provided communication, security, and delivery responsibilities are clearly defined.</p><p>In short, the objective is not simply to find the <strong>lowest-cost provider.</strong> It is to find a partner that can deliver the required capability with the right level of quality, accountability, and long-term maintainability.</p><p>If you’re exploring what’s <strong>next for SAP BTP in your organization</strong>, it may be worth discussing where these capabilities could create the most value and what a practical path forward might look like. You can<strong> </strong><a href="https://inclusioncloud.com/contact-us/"><strong>schedule a quick call</strong></a> with our team to explore the possibilities with one of our specialists.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=79db49bc961f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Best Nearshore SAP and AI Consulting Firms]]></title>
            <link>https://medium.com/@inclusion.cloud/best-nearshore-sap-and-ai-consulting-firms-29c0b72a47a1?source=rss-aa686e488265------2</link>
            <guid isPermaLink="false">https://medium.com/p/29c0b72a47a1</guid>
            <category><![CDATA[outsourcing]]></category>
            <category><![CDATA[nearshore]]></category>
            <category><![CDATA[consulting]]></category>
            <category><![CDATA[sap]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Inclusion Cloud ]]></dc:creator>
            <pubDate>Thu, 15 Oct 2026 13:41:57 GMT</pubDate>
            <atom:updated>2026-10-16T13:46:11.078Z</atom:updated>
            <content:encoded><![CDATA[<h3>What Works, What Doesn’t, and What to Look For</h3><p>Nearshoring has been around long enough that most of its advantages no longer need much introduction.</p><p>In simple terms, it gives U.S. companies access to new pools of remote technical resources that work within nearly the <strong>same time zones as their internal teams</strong>, often at more competitive rates. That makes distributed delivery easier without introducing the communication gaps that usually come with larger time-zone differences.</p><p>Those advantages are still valuable for many companies that need to scale their teams. But if you are choosing a partner today for projects <strong>around AI and agents</strong>, there may be other engagement models that are better suited to the way this kind of work is now being done than traditional <a href="https://inclusioncloud.com/services/staff-augmentation/">staff augmentation</a> or <a href="https://inclusioncloud.com/services/dedicated-teams/">dedicated teams</a>.</p><p>The reason is fairly simple: the way developers and engineers ship software is not the same as it was five years ago.</p><p>Teams can now produce more in less time, with AI taking on a growing share of execution. Engineers are still in control, but more of their time is shifting toward review, architecture, context, and decision-making.</p><p>This is already happening inside some of the biggest tech companies.</p><p><a href="https://www.cnbc.com/2025/04/29/satya-nadella-says-as-much-as-30percent-of-microsoft-code-is-written-by-ai.html">Satya Nadella has said</a> that roughly <strong>20% to 30% of the code in some Microsoft repositories and projects is already written by AI</strong>, and that share is growing. Google has reported similar numbers, while Shopify has gone as far as asking teams to show why AI cannot handle a need before requesting more headcount.</p><p>The important takeaway is not that engineers are going away. It is that the way their time is being used is changing.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/901/0*sROeAj_w3yPeqLJW.png" /></figure><h3>What this means for consulting</h3><p>The traditional nearshore model still tends to be built around <strong>people and hours</strong>. A client needs more capacity, a vendor adds resources, and the engagement scales largely with the amount of time those people spend on the project.</p><p>For many types of work, that remains perfectly reasonable.</p><p>AI projects, however, introduce a different dynamic.</p><p>A company might be building an AI product from scratch, creating an agent with Claude, automating part of a business process, or trying to make better use of the AI capabilities now appearing inside big platforms such as <strong>SAP or Oracle</strong>.</p><p>The details can change depending on the project, but the idea is usually the same: automate more of the work, reduce repetitive tasks, and help developers get more done in less time. In the end, it comes down to productivity.</p><p>This is where we start to see some friction with the hourly billing model.</p><p>If better tooling, <a href="https://inclusioncloud.com/services/ai-accelerators/">reusable accelerators</a>, or agents allow a team to complete in 500 hours what might previously have required 1,000, that is a huge productivity gain. The client may benefit from fewer hours billed, but the commercial model itself is still tied to time consumed rather than to what was accomplished.</p><p>That is where <strong>outcome-based delivery</strong> starts to become more interesting.</p><p>Instead of buying a block of hours, the client buys a defined piece of work: a workflow implemented, an agent deployed, a milestone reached, or a capability delivered.</p><p>The incentives shift with it.</p><p>If the provider finds a better way to execute the work, the client gets the result sooner and the team can move on to the next initiative. Automation stops being something that simply compresses billable time and becomes part of the delivery model itself.</p><p>It can be a <strong>win-win</strong> for both sides: the client gets the desired outcome faster, potentially at a lower overall cost, while the partner can work with a leaner, more focused team.</p><p>For AI projects, that feels like a much more natural fit.</p><h3>1. What Works: The Nearshore Advantage Is Still There</h3><p>There is a reason nearshoring became such a common delivery model in the first place.</p><p>Working in roughly the same time zone makes day-to-day collaboration easier, particularly when projects involve several business and technical stakeholders. A SAP engineer can join a call with finance. An integration specialist can work through an issue with a U.S.-based architect in real time. Requirements can be clarified <strong>during the same working day</strong> rather than disappearing into an overnight handoff.</p><p>That matters for global companies, where the difficult part is often not writing code but understanding how a process actually works across teams, systems, approvals, data, and exceptions.</p><p>Latin America (or LATAM) also offers a broad pool of resources across software engineering, SAP, Oracle, cloud, data, integrations, and AI. English proficiency, experience with U.S. organizations, and cultural proximity all help external teams operate more like an extension of the client organization.</p><p>There is also room to mix delivery models. Some initiatives benefit from having <strong>certain roles onshore</strong> while the broader engineering work happens nearshore. The point is not to force every role into the same geography, but to assemble the right combination for the project.</p><p>That flexibility remains one of nearshoring’s biggest strengths.</p><h3>Big Consulting vs. Boutique Nearshore Partners: What Actually Changes?</h3><p>Not all nearshore consulting engagements offer the same experience. And when it comes to choosing a partner, <strong>size can make a difference</strong>.</p><p>Large consulting firms bring global resources, established delivery structures, and the capacity to manage complex SAP and AI transformations across multiple countries and business units.</p><p>But there’s also a strong case for working with a <strong>boutique firm</strong>, especially when you have a specific project in mind.</p><p>Boutique partners typically offer a more direct, less bureaucratic way of working. With smaller teams and fewer layers of management, service managers can maintain closer relationships with both the consultants doing the work and the client’s internal team. This means faster communication, more direct access to senior engineers, greater flexibility, and less time spent on coordination.</p><p><strong>1. Accenture | Global Consulting</strong></p><ul><li><strong>Key strengths:</strong> SAP transformations, enterprise AI, global delivery capacity.</li><li><strong>Best suited for:</strong> Large, complex transformation programs.</li></ul><p><strong>2. TCS | Global IT Services</strong></p><ul><li><strong>Key strengths:</strong> SAP, offshore and nearshore delivery, AI platforms.</li><li><strong>Best suited for:</strong> Enterprise-scale implementations and ongoing operations.</li></ul><p><strong>3. Globant | Digital Engineering</strong></p><ul><li><strong>Key strengths:</strong> AI-native development, AI Pods, distributed engineering.</li><li><strong>Best suited for:</strong> Digital products and AI engineering.</li></ul><p><strong>4. Infosys | Global IT Consulting</strong></p><ul><li><strong>Key strengths:</strong> SAP consulting, Infosys Topaz, AI-powered transformation.</li><li><strong>Best suited for:</strong> Enterprise modernization, SAP implementations, and AI adoption.</li></ul><p><strong>5. Inclusion Cloud | Boutique Enterprise Technology Partner</strong></p><ul><li><strong>Key strengths:</strong> SAP, Oracle, certified engineers, AI Pods, and accelerators.</li><li><strong>Best suited for:</strong> Flexible enterprise AI delivery, platform optimization, and team scaling.</li></ul><h3>2. What Doesn’t: Scaling Delivery Only by Adding More Hours</h3><p>In the traditional model, the relationship is fairly direct: more work usually means more people, which means more billable hours.</p><p>AI is starting to <strong>weaken</strong> that connection.</p><p><a href="https://www.anthropic.com/institute/recursive-self-improvement">Anthropic offers a pretty striking example</a>. In the second quarter of 2026, <strong>the typical engineer there was merging 8x as much code per day as in 2024</strong>, with more than 80% of the code merged into its codebase authored by Claude. Of course, the same report cautions that lines of code are not a perfect measure of productivity. But still, it gives a pretty clear picture of where <a href="https://inclusioncloud.com/insights/blog/tech-hiring-ai-method-inmove/">software development is heading</a>.</p><p>That is also changing where their time goes. As more of the execution is handled by AI, engineering work shifts further toward architecture, enterprise context, quality, security, business rules, edge cases, and judgment. In fact, Anthropic says human <strong>code review is already becoming a new bottleneck</strong> as AI increases the amount of code moving through the organization.</p><p>The result is a delivery unit that can potentially handle much more without growing at the same pace.</p><p>And this is where hourly billing can start to feel misaligned. If the delivery system becomes significantly more productive, the commercial model should be able to recognize that gain rather than continue measuring value primarily through time consumed.</p><h3>How Are Leading Consulting Firms Adapting Their AI Engagement Models?</h3><p>AI is changing more than the technologies consulting firms implement. It’s also changing how they structure delivery, organize engineering teams, and charge for their services.</p><p>Here are five examples of how consulting providers are approaching this shift.</p><p><strong>1. Accenture | AI Refinery &amp; AI Factories</strong></p><ul><li><strong>Engagement approach:</strong> Enterprise AI transformation and platform-led delivery.</li><li><strong>Best suited for:</strong> Organizations deploying and governing AI across complex enterprise environments.</li></ul><p><strong>2. TCS | AI WisdomNext &amp; Enterprise AI Factories</strong></p><ul><li><strong>Engagement approach:</strong> AI Factory implementations and large-scale enterprise transformation programs.</li><li><strong>Best suited for:</strong> Organizations building shared infrastructure and capabilities to scale AI across the business.</li></ul><p><strong>3. Globant | AI Pods &amp; Glob.AI</strong></p><ul><li><strong>Engagement approach:</strong> Subscription and consumption-based AI services, with agents supervised by specialists and pricing aligned to outputs.</li><li><strong>Best suited for:</strong> AI-powered software engineering and repeatable business workflows.</li></ul><p><strong>4. Infosys | Infosys Topaz</strong></p><ul><li><strong>Engagement approach:</strong> AI-first consulting and delivery combining enterprise AI services, reusable solutions, and AI-powered engineering.</li><li><strong>Best suited for:</strong> Large-scale AI adoption, enterprise modernization, and intelligent automation.</li></ul><p><strong>5. Inclusion Cloud | AI Pods (NEW) &amp; Enterprise AI Accelerators</strong></p><ul><li><strong>Engagement approach:</strong> Lean teams of senior AI-native engineers combining agentic workflows, reusable accelerators, and human supervision to deliver defined outcomes.</li><li><strong>Best suited for:</strong> Enterprise AI initiatives, SAP and Oracle workflows, AI product development, and moving pilots into production.</li></ul><h3>3. What to Look For in an AI-Ready Nearshore Partner</h3><h4>Where AI Pods Fit</h4><p>This is the thinking behind our <a href="https://inclusioncloud.com/services/ai-pods/">AI Pods</a> at Inclusion Cloud.</p><p>An AI Pod is a <strong>compact team of senior specialists working with AI agents</strong> and reusable accelerators around a defined initiative.</p><p>Senior engineers set the direction and design how the agents will work. That includes looking closely at the economics of the system through an <a href="https://inclusioncloud.com/insights/blog/ai-finops-autonomous-ai/">AI FinOps lens</a>: choosing the right models when that decision has not already been made through procurement, monitoring usage and token costs, and reviewing the workflow itself.</p><p>That last part matters more than it may seem. An inefficient process does not become efficient just because AI is added to it. In many cases, those inefficiencies simply show up later in the monthly AI bill.</p><p>Once the workflow is defined, agents handle more of the execution. Domain specialists then supervise the outputs, review exceptions, and help identify where the same approach could be extended to other processes if the client wants to keep scaling.</p><p>The idea is to keep the team lean, but not for the sake of simply having fewer people. The goal is to create a delivery unit with enough seniority and context to own a meaningful piece of work without adding unnecessary coordination or execution overhead.</p><p>This structure also fits naturally with the <strong>outcome-based model </strong>we discussed earlier. As the Pod improves the workflow, automates more of the execution, or finds more efficient ways to use models and agents, those gains can translate into faster delivery and better economics for the client.</p><h4>And What Happens When the Project Is Finished?</h4><p>This is another area we think buyers should pay attention to.</p><p>A consulting engagement should leave the client with something useful after the team is gone.</p><p>If the work produces an agent, a workflow, a product, or another enterprise capability, the client should be able to own it and continue from there.</p><p>That leaves two sensible paths.</p><ul><li>One is to keep the Pod and apply the model to another process.</li><li>The other is to take what has been built and continue internally.</li></ul><p>The important part is that the <strong>choice belongs to the client</strong>.</p><h3>So What Makes a Strong Nearshore SAP and AI Partner?</h3><p>The answer is becoming broader than geography or hourly rates.</p><p>Time-zone alignment still matters. So do strong resources, cultural proximity, enterprise platform expertise, and the flexibility to combine nearshore and onshore delivery.</p><p>But for AI projects, there is another layer.</p><p>The partner should be able to <strong>use AI to improve the delivery system itself</strong>.</p><p>That means knowing where agents can execute, where engineers need to stay in control, how to reuse what already works, how to measure what is being delivered, and how to structure the engagement so productivity gains are useful to both sides.</p><p>That is the direction we see nearshoring moving.</p><p>The fundamentals are still good.</p><p>The delivery model is simply starting to catch up with the technology.</p><p>And if you want to explore the <strong>AI opportunities</strong> inside your company, learn more about <a href="https://inclusioncloud.com/#:~:text=inMOVE%E2%84%A2%20by%20InclusionCloud"><strong>inMOVE by Inclusion Cloud</strong></a>, our AI-driven recruiting method for identifying top consultants in the market, or discuss where one of these models could fit, you can <a href="https://inclusioncloud.com/contact-us/">schedule a quick call</a> with one of our specialists.</p><p><em>Originally published at </em><a href="https://inclusioncloud.com/insights/blog/best-nearshore-sap-and-ai-firms/"><em>https://inclusioncloud.com</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=29c0b72a47a1" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[SAP Connect 2026: Key Takeaways From the Keynote]]></title>
            <link>https://medium.com/@inclusion.cloud/sap-connect-2026-key-takeaways-from-the-keynote-e171b8bbded2?source=rss-aa686e488265------2</link>
            <guid isPermaLink="false">https://medium.com/p/e171b8bbded2</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[llm]]></category>
            <category><![CDATA[sap]]></category>
            <category><![CDATA[agents]]></category>
            <dc:creator><![CDATA[Inclusion Cloud ]]></dc:creator>
            <pubDate>Wed, 07 Oct 2026 14:15:28 GMT</pubDate>
            <atom:updated>2026-10-07T15:47:36.907Z</atom:updated>
            <content:encoded><![CDATA[<p>SAP Connect 2026 in Las Vegas offered a pretty clear look at where SAP believes enterprise AI is heading next: toward more autonomous operations, with agents working across finance, procurement, supply chain, HR, customer experience, and other core business functions.</p><p>But getting there raises a harder question.</p><p>AI agents are getting better fast.</p><p>Yet if they are going to start touching critical parts of the business, “better” is not really the standard.</p><p>It has to be right. And it has to be right consistently, too.</p><p>That tension ran through much of the keynote.</p><p>SAP spent the session showing what it calls the <strong>Autonomous Enterprise</strong>, with agents working across key business functions.</p><p>And there was a clear thread running through all of it: how do you give agents enough context to make the right decisions, and enough control to trust them with more autonomy?</p><p>Christian Klein, SAP CEO, framed the first part early in the keynote. Frontier models are getting better, but they still “don’t know how your business works.” Ask one which of your suppliers will deliver next week and, without that business context, it will guess.</p><p>Which raises the first question:</p><p><strong>If LLMs are probabilistic, how do you get the right answer every time?</strong></p><p>Then SAP Chief Security Officer Marielle Ehrmann brought up the other side of the problem. As agents move from answering questions to taking actions inside critical processes, reliability is not enough. You also need control.</p><p>She summed it up with another question:</p><p><strong>“How fast can we deploy AI without making tomorrow’s headlines?”</strong></p><p>Those two questions capture much of the logic behind what SAP showed at Connect: give agents the business context they need to make reliable decisions, then add the governance, traceability, and human oversight required to safely increase autonomy.</p><p>Here’s a closer look at the main takeaways from SAP Connect 2026.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*-aB4QBWqoJE7f7yC.png" /></figure><h3>A Good Answer Is Not Enough</h3><p>One of SAP’s clearest messages was that enterprise AI cannot depend on the frontier models alone.</p><p>LLMs are probabilistic by nature. They can reason, interpret language, generate recommendations, and produce answers that sound convincing.</p><p>But sounding right and being right are not the same thing. SAP made that distinction explicit during the keynote: for many enterprise tasks, <strong>“a good answer is not enough.”</strong></p><p>If an agent is involved in moving money, making a procurement decision, or taking action inside an operational process, the result also needs to be correct, compliant, and traceable.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*rD3qnaFGaKgdtgl5.png" /></figure><h3>SAP is not trying to make the LLM deterministic</h3><p>Giving an LLM more context does not suddenly make the model deterministic.</p><p>LLMs still generate outputs probabilistically, predicting the next token based on the information available to them.</p><p>SAP’s approach is to <strong>reduce the uncertainty</strong> around those outputs by grounding agents in the business context they need to run.</p><p>That means giving them access to the same data, business rules, permissions, processes, and relationships that already define how the company works.</p><p>So when an agent moves into a real business process, it is not free to “invent” how that process should run.</p><h3>Business Context Is Where SAP Thinks It Has an Advantage</h3><p>This explains why <strong>context</strong> came up so often during SAP Connect.</p><p>A generic model may know what a purchase order is.</p><p>It does not automatically know your suppliers, approval thresholds, inventory position, company policies, permissions, contracts, master data, or how one business event connects to another.</p><p>That business context is what SAP wants its agents to understand.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*0fKXRfOaU0ldNIdA.png" /></figure><h3>Business AI Platform: Build, Contextualize, Govern</h3><p>SAP positioned the <a href="https://www.sap.com/products/ai-platform.html"><strong>Business AI Platform</strong></a> as the layer where companies can build, contextualize, and govern AI across their landscape.</p><p><a href="https://www.sap.com/products/artificial-intelligence/joule-studio.html">Joule Studio</a> handles the build side, allowing companies to extend SAP agents or create their own.</p><p>Then Business Data Cloud and Knowledge Graph provide the context.</p><p>SAP described this layer as a way for agents to understand business data, processes, policies, relationships, and logic across SAP and third-party environments.</p><p><strong>Knowledge Graph</strong>, in particular, was presented as the map that helps agents navigate processes, data flows, business objects, relationships, and rules.</p><p>Instead of explaining the company from scratch every time, the goal is for the agent to work with a structured understanding of how the business actually works.</p><h3>Joule Work Becomes the AI Engagement Layer</h3><p>That business context comes together in <strong>Joule Work</strong>, which SAP presented as a new AI engagement layer across the enterprise.</p><p>Christian Klein summed it up with a simple phrase: <strong>“a new UI for AI.”</strong></p><p>The idea is pretty straightforward.</p><p>For years, users had to learn how SAP worked: transaction codes, exports to spreadsheets, presentations, and a lot of manual steps in between.</p><p>With Joule Work, SAP wants that interaction to feel much more natural.</p><p>You can talk to SAP in your own language, ask questions, delegate tasks, analyze SAP and non-SAP information, and let Joule connect you with the right assistant or agent.</p><p>SAP organized the experience around three areas: <strong>Conversations</strong>, for asking questions and delegating work; <strong>Business Insights</strong>, for analyzing and visualizing business information; and <strong>Joule Studio</strong>, where companies can extend existing SAP agents or build their own.</p><p>The important part is that Joule is not just another chat interface sitting on top of the system.</p><p>It is the place where users can interact with the data, business context, agents, and workflows already running underneath.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*haG99JPe2U-8wZK2.png" /></figure><h3>110,000 SAP employees are already using Joule</h3><p>SAP also brought its own internal usage numbers to the stage.</p><p>According to the keynote, around <strong>110,000 SAP employees</strong> are already using <a href="https://inclusioncloud.com/insights/control-joule-ai-unit-spending-sap/">Joule</a> to perform hundreds of thousands of tasks every day.</p><p>SAP reported productivity gains of more than <strong>20%</strong> across areas including HR, finance, procurement, and supply chain.</p><p>Klein used earnings preparation as one example.</p><p>What used to take weeks of coordination across finance, investor relations, and communications can now be supported by Joule using internal financial information and external analyst data, surfacing relevant KPIs and helping anticipate questions.</p><p>The larger point is that SAP does not want Joule to be just another chatbot.</p><p>It wants it to become the layer through which users increasingly delegate work across the business.</p><h3>From Individual Agents to an Autonomous Suite</h3><p>SAP also made clear that the conversation is moving beyond individual agents.</p><p>The company said it has more than <strong>20 assistants and over 200 agents generally available</strong>, with a target of more than <strong>400 agents by the end of the year</strong>.</p><p>Those assistants and agents are being organized across major business domains including finance, spend, HCM, supply chain, customer experience, and industry AI.</p><p>But the more important idea is orchestration.</p><p>A single agent does not need to know how to do everything.</p><p>The assistant can identify the task, select the right specialized agent, connect it to the relevant context, and move the work through the right business process.</p><p>That is a very different model from simply adding a chatbot on top of an application.</p><h3>Procurement: More Analysis, Less Manual Coordination</h3><p>Procurement offered one of the clearest examples of how SAP sees agents being applied in practice.</p><p>A customer described the challenge of managing hundreds of categories and suppliers while creating RFPs, analyzing bids, and preparing negotiations.</p><p>At scale, those processes take time and can become inconsistent, particularly in highly regulated environments where traceability and auditability matter.</p><p>The company has been piloting SAP’s sourcing assistant around bid analysis and negotiation support.</p><p>The goal is for agents to handle more of the analytical heavy lifting: creating bids, comparing offers, identifying negotiation levers, and bringing more consistency to work that previously required significant manual effort and coordination.</p><p>The customer said it expects a <strong>25% reduction in cycle times and a 35% productivity uplift</strong>.</p><p>But the advice around deployment was just as important as the expected gains.</p><ul><li>Start with reliable data.</li><li>Choose processes where a wrong decision is recoverable.</li><li>Involve the people who know where the actual friction is.</li><li>And do not underestimate change management.</li></ul><p>That message came back several times during the keynote.</p><p>Automation does not compensate for weak foundations.</p><h3>Supply Chain: From Reactive to Predictive</h3><p>Morgan Foods gave one of the strongest examples of what those foundations can enable.</p><p>Before its SAP modernization, much of its supply chain planning was manual, siloed, and dependent on Excel.</p><p>With S/4HANA and IBP, the company connected more of the operation across forecasting, ingredients, production scheduling, warehousing, distribution, and supplier collaboration.</p><h3>From 3–4 weeks to about 5 minutes</h3><p>Morgan Foods, one of the companies invited to share their case at the stage, said it once took <strong>three or four weeks</strong> to determine whether it could fulfill a large customer request.</p><p>Today, that can be answered in about <strong>five minutes</strong>.</p><p>The important part is how they got there.</p><p>This was not the result of simply putting an agent on top of fragmented systems.</p><p>The company first built a more connected operational foundation.</p><p>Now that same foundation can support the next shift.</p><p>Morgan Foods CFO Steve Henke described it as moving <strong>from reactive to predictive</strong>.</p><p>Agents can connect demand signals, inventory levels, and supplier status much faster than a person or team could.</p><p>But Henke was also direct about the prerequisites: the right data, good master data, and clean, documented business processes.</p><p>And people still matter.</p><p>Henke emphasized that humans need to remain above the agents because AI cannot replace judgment and experience.</p><h3>Autonomy Starts Connecting Functions</h3><p>This is where SAP’s Autonomous Enterprise narrative becomes more interesting.</p><p>The goal is not just to automate departments individually.</p><p>SAP wants an event in one part of the business to trigger coordinated actions across several others.</p><p>One keynote scenario used an acquisition as the example.</p><p>Bringing new employees into the company was not treated as an isolated HR task.</p><p>It also affected IT, facilities, procurement, access, devices, and supplier availability.</p><p>When laptops were at risk of arriving late, the workflow could identify the problem, look for an alternative supplier, route the necessary approval, and continue the onboarding process.</p><p>That is much closer to how real business processes behave.</p><p>They cross functional boundaries.</p><p>And that is exactly where context becomes more important.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*UTxpaEZecXm0UVHH.png" /></figure><h3>TechWolf Adds More Workforce Context</h3><p>SAP also announced the acquisition of <strong>TechWolf</strong>, expanding its capabilities around skills and workforce intelligence.</p><p>The idea is to improve visibility into which skills already exist inside an organization, where gaps remain, how roles are changing, and where people could develop or move next.</p><p>SAP said it expects TechWolf intelligence to be integrated into SuccessFactors across workforce planning, learning, mobility, recruiting, and the broader SAP ecosystem.</p><p>Again, the pattern is consistent.</p><p>More context gives the system more information about what already exists before another decision gets made.</p><p>Before hiring externally, leaders can better understand internal capabilities.</p><p>Once someone joins, the same context can help identify development paths, learning opportunities, and future roles.</p><h3>SAP Pay Brings Agents Closer to Execution</h3><p>Finance also brought one of the event’s most concrete announcements.</p><p>Christian Klein introduced <strong>SAP Pay</strong>, a payment service embedded directly into SAP Cloud ERP.</p><p>He framed the problem in familiar terms.</p><p>Paying a supplier can mean approving an invoice in one system, initiating a bank transfer somewhere else, and then reconciling everything afterwards.</p><p>Three separate steps for something that, in Klein’s words, should simply happen.</p><p>SAP Pay is designed to bring those steps into the same flow.</p><p>Once an invoice is approved, SAP agents can execute the payment and reconcile it automatically inside the ERP. SAP also said the service will support several payment methods, including cross-border payments and digital currencies.</p><p>The company also pointed to the potential for around a <strong>25% TCO reduction on average</strong>.</p><p>This matters because agents are moving beyond recommendations.</p><p>They are getting closer to execution.</p><p>And that makes the second question from the keynote much more important.</p><h3>How Fast Can We Deploy AI Without Making Tomorrow’s Headlines?</h3><p>SAP Chief Security Officer Marielle Ehrmann put the issue directly.</p><p>A year ago, she said, boardrooms were asking:</p><p><strong>How fast can we deploy AI?</strong></p><p>Now the question has shifted:</p><p><strong>“How fast can we deploy AI without making tomorrow’s headlines?”</strong></p><p>That is the other side of autonomy.</p><p>Once an agent can take action in a process, access business data, interact with systems, or execute a transaction, you need much more than a capable model.</p><ul><li>You need to know who owns it.</li><li>What it can access.</li><li>What actions it is allowed to take.</li><li>What level of risk it carries.</li><li>And who can step in if something goes wrong.</li></ul><h3>SAP Calls It “Freedom Within Boundaries”</h3><p>Eman described SAP’s approach as <strong>“freedom within boundaries.”</strong></p><p>Agents operate within identity and access controls.</p><p>They inherit the authorizations of the person they represent and can only access data and perform actions they are explicitly permitted to execute.</p><p>SAP also emphasized governance and visibility.</p><p>As companies deploy hundreds of agents across SAP and non-SAP environments, they need to know who owns each one, what it can access, and what risk it carries.</p><p>In this model, governance is not a passive compliance checklist added at the end.</p><p>It becomes part of how the agent operates.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*jyiaxHwSmfiZMF6H.png" /></figure><h3>What Does It Take to Get There?</h3><p>For all the talk about agents and autonomy, SAP’s answer came back to some fairly traditional foundations.</p><ol><li>You need good data.</li><li>You need good processes.</li><li>And you need people to change how they work.</li></ol><h4>1. Get the Data Right</h4><p>If the underlying data is fragmented or unreliable, giving an agent access to more of it does not solve the problem.</p><p>Several speakers returned to trusted data, master data, harmonization, and unified business context.</p><p>One customer explained that when data is fragmented across legacy systems, there is very little for AI to work with.</p><p>AI starts becoming useful at scale when the underlying data is harmonized, trustworthy, connected, and prepared for it.</p><h4>2. Get the Processes Right</h4><p>Agents also need clarity around how work is supposed to happen.</p><p>That means documented processes, defined approval paths, business rules, and a clear understanding of when people need to intervene.</p><p>Morgan Foods put it simply: technology delivers when the foundation is there, including the right data, master data, and clean business processes.</p><h4>3. Do Not Underestimate Change Management</h4><p>And then there is the human side.</p><p>Christian Klein returned to this near the end of the keynote.</p><p>AI, he said, requires significant change management if companies want it to deliver the expected outcomes.</p><p>Based on his own experience driving transformation inside SAP, he described change management as one of the hardest parts of the job.</p><p>That may be one of the easiest points to overlook.</p><p>You can have a technically strong agent and still fail if people do not trust it, understand how to use it, know where they remain accountable, or change the surrounding process.</p><h3>The Road From Semi-Autonomous to Autonomous</h3><p>Two questions probably summarize the keynote better than anything else:</p><ol><li><strong>Can an agent give you the right answer consistently?</strong></li><li><strong>Can you trust it enough to give it more autonomy?</strong></li></ol><p>Those questions matter because CIOs around the world are trying to answer them inside their own environments.</p><p>And as SAP made clear throughout the keynote, the answer starts with the foundations.</p><p>The data has to be ready. The processes have to be clean and optimized. And the business context needs to be structured in a way that agents can actually use.</p><p>Then comes change management.</p><p>Teams will need to adapt to new ways of working, where people increasingly supervise, review, and guide the process while agents take on more of the execution.</p><p>That is where many companies are today: in a semi-autonomous stage.</p><p>SAP’s vision is that agents will gradually take on more complex work, moving from assistance to execution and helping compress processes that once took weeks or months into days, minutes, or even seconds.</p><p>But that level of autonomy only works if the foundations are ready.</p><p>At Inclusion Cloud, we have been working with SAP for more than 20 years. If you are looking to accelerate your agent implementation with the right business context, we can help with certified SAP, data, and AI resources. <a href="https://inclusioncloud.com/technologies/sap-consulting-and-solutions/">Book a call with us!</a></p><p><em>Originally published at </em><a href="https://inclusioncloud.com/insights/blog/sap-connect-2026-ai-agents-reliable/"><em>https://inclusioncloud.com</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e171b8bbded2" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Exception Queue Could Be Killing Your Agent ROI]]></title>
            <link>https://medium.com/@inclusion.cloud/the-exception-queue-could-be-killing-your-agent-roi-9b47a1b9c372?source=rss-aa686e488265------2</link>
            <guid isPermaLink="false">https://medium.com/p/9b47a1b9c372</guid>
            <category><![CDATA[agentic-ai]]></category>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[llm]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Inclusion Cloud ]]></dc:creator>
            <pubDate>Fri, 25 Sep 2026 15:37:14 GMT</pubDate>
            <atom:updated>2026-09-25T15:37:14.474Z</atom:updated>
            <content:encoded><![CDATA[<p>A client we had worked with in the past recently came back to us looking for advice on an AI agent workflow they had implemented.</p><p>The agents were handling roughly <strong>80% of the standard cases without human intervention</strong>, which looked promising at first.</p><p>But the remaining 20% was starting to change the economics.</p><p>Those cases were being pushed into an exception queue, where people had to review outputs, resolve ambiguities, and correct edge cases. As the queue grew, so did the <strong>amount of human work </strong>around the automation.</p><p>Some of the savings they had gained by automating the process were <strong>starting to disappear</strong> into exception handling.</p><p>What they wanted help thinking through was why so many cases were reaching the queue in the first place, and whether some of that human review could be avoided.</p><p>Not every case required real human judgment. Some of the human review was not there because the case was inherently complex. It was there because the <strong>uncertainty had never been resolved upstream</strong>.</p><p>That 80/20 split is anecdotal, not an industry benchmark. But it raises a much broader question:</p><p>How much human intervention can an AI workflow absorb before it starts undercutting the savings it was supposed to create?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*UbDYnhYogRT_ULbAbahsvw.jpeg" /></figure><h3>Human-In-the-Loop Has a Cost</h3><p><a href="https://hai.stanford.edu/ai-definitions/what-is-human-in-the-loop">Human-in-the-loop, or HITL,</a> is an important part of many agent architectures.</p><p>There are decisions we should not fully delegate. High-risk actions, unusual customer situations, regulatory requirements, irreversible transactions, and genuinely ambiguous cases may all justify human review.</p><p>The problem starts when too many cases routinely require human intervention.</p><p>Every review adds labor. It can also add latency, create queues, and limit how quickly the workflow can scale.</p><p>And if we want to understand <strong>what it really costs to automate a business process with agents</strong>, all of that needs to be part of the math.</p><p>The calculation is not simply:</p><p><strong>Human cost vs. Agent cost</strong></p><p>A more realistic comparison is:</p><p><strong>Cost of the current process vs. Cost of the agentic workflow</strong></p><p>And that agentic workflow may include:</p><p><strong>Inference + retries + exception handling + human review + integrations + monitoring + maintenance</strong></p><p>That is the number that ultimately tells us whether automating the process with agents actually makes economic sense.</p><p><a href="https://www.cloudzero.com/blog/human-in-the-loop-cost/">CloudZero makes a similar argument</a> in its analysis of HITL economics: pricing the agent alone can significantly understate the cost of the production system because reviewer labor and operational overhead remain part of the workflow.</p><p>There is another problem. As the review queue grows, the business may need more reviewers to preserve the same level of scrutiny.</p><p>The agent scales like software.</p><p>The supervision layer can start scaling like <strong>headcount</strong>.</p><h3>But Not Every Exception Is Really an AI Problem</h3><p>Before adding more reviewers, there is a more useful question to ask:</p><p><strong>Why is the agent escalating the case in the first place?</strong></p><p>Some exceptions legitimately require judgment.</p><p>Imagine a procurement agent processing purchase requests. A request significantly above normal thresholds, involving an unusual supplier or creating a compliance risk, may deserve human approval.</p><p>But other exceptions look very different.</p><ul><li>Two systems contain different addresses for the same customer.</li><li>Finance and Sales use different definitions of an active customer.</li><li>The agent retrieves two versions of the same policy.</li><li>An old document ranks above the current one.</li><li>The database contains the data, but the business definition needed to interpret it exists only in someone’s head.</li></ul><p>These cases may end up in a human review queue, but they are not necessarily failures of AI reasoning.</p><p>They are often <strong>data problems</strong> appearing downstream as HITL problems.</p><h3>When Poor Data Turns Humans into the Resolution Layer</h3><p>A <a href="https://www.cio.com/article/4215442/when-ais-human-in-the-loop-really-isnt.html?utm_source=chatgpt.com">recent CIO.com</a> article gives a good example of how this can happen.</p><p>In this case, an agent system was pulling customer information from several APIs. The same customer field could come back with different values depending on the source, exposing the agent to <strong>semantic ambiguity</strong>.</p><p>In simple terms, <a href="https://www.gartner.com/en/newsroom/press-releases/2026-05-11-gartner-says-lack-of-semantics-causes-inaccurate-artificial-intelligence-agents-and-wasted-spending">semantic ambiguity</a> appears when the data is available, but its business meaning has not been fully defined. For example, if one source contains a billing address, another the most recently verified address, and another the address used in the latest transaction, what exactly should <strong>“the customer’s address”</strong> mean for this particular task?</p><p>Without a business rule defining that meaning and identifying which value should be treated as <strong>authoritative</strong>, the agent is left to interpret it on its own.</p><p>At first, the agents were expected to resolve that ambiguity themselves.</p><p>That didn’t work consistently.</p><p>So the team moved the decision upstream. They added business rules to reconcile the records before the data ever reached the agents, giving them one approved value to work with.</p><p>It sounds like a small architectural change, but in production it makes a big difference.</p><p>Without resolution:</p><p><strong>Multiple values → Agent interprets → Uncertainty or error → Human reviews</strong></p><p>With resolution:</p><p><strong>Multiple values → Business rule resolves ambiguity → Agent receives trusted data</strong></p><p>The second approach removes an entire category of potential exceptions.</p><p>And that matters for more than just reducing the headcount needed for human review.</p><p>Human attention is limited. The more exceptions reviewers have to work through, especially repetitive ones, the harder it becomes to give each case the<strong> same level of attention</strong>.</p><p>That creates two problems.</p><p>First, reviewing hundreds of routine cases can create fatigue. The reviewer becomes used to seeing the same patterns over and over, making it easier for the genuinely unusual case to slip through.</p><p>Second, highly accurate AI can create a different kind of problem: <strong>the</strong> <strong>reviewer starts expecting the AI to be right</strong>.</p><p>Eric Billingsley, COO and CTO of TrustScale, makes this point in the CIO article. When a system is correct most of the time, reviewers can gradually shift from independently evaluating each decision to simply confirming what the system has already suggested.</p><p>After hundreds of correct recommendations, the rare mistake becomes easier to approve automatically.</p><h3>Trusted Data Can Shrink the Exception Queue</h3><p>When it comes to data, there is no single “trusted data layer” that solves every agent problem.</p><p>Different types of ambiguity require different mechanisms.</p><ul><li><strong>Data reconciliation</strong> can determine which value should be used when several applications return different information for the same field.</li><li><strong>Semantic layers</strong> can encode agreed business definitions for metrics and entities such as ARR, margin, churn, customer status, or qualified opportunity.</li><li><strong>Governed retrieval</strong> can establish which documents take precedence based on source, version, customer, geography, or recency.</li><li><strong>Metadata</strong> and data contracts can make ownership, meaning, freshness, and expected quality explicit before the agent acts on the data.</li></ul><p>The implementation changes depending on the problem.</p><p>The objective does not:</p><p><strong>Resolve predictable ambiguity before the agent encounters it.</strong></p><p>That can have a direct impact on the <a href="https://inclusioncloud.com/insights/blog/ai-finops-autonomous-ai/">economics </a>of HITL.</p><p>A stronger data architecture can reduce avoidable exceptions, so human review is reserved for cases that actually require judgment, rather than becoming the fallback when the agent lacks context or a clear rule.</p><p>That means less review overhead, less latency, and more human attention available for the cases where it actually matters.</p><h3>Human intervention rate should become an AI FinOps metric</h3><p>Most conversations about AI costs still start with models and token consuption.</p><ul><li>How many tokens did we consume?</li><li>Which model handled the request?</li><li>How much did inference cost?</li></ul><p>Those metrics are important, but they are not enough to get a realistic sense of what it will actually cost to implement an agent in a business process. And that is a fundamental part of putting AI into production.</p><p>One of the biggest risks is successfully automating a process, only to realize later that the automation costs more to operate than the previous workflow, when there were no agents involved and people were simply handling the work themselves.</p><p>That is why we also need to account for <strong>everything around the agent</strong>: the people involved, potential bottlenecks, delays, exceptions, monitoring, and maintenance.</p><p>For example, if a $0.50 execution creates ten minutes of work for an expensive specialist, looking at the inference cost alone gives us a very incomplete picture of the economics.</p><p>Teams operating agents at scale should also start measuring things like:</p><ul><li><strong>Human Intervention Rate: </strong>What percentage of executions require a person?</li><li><strong>Exception Rate:</strong> How often does the agent leave the expected path?</li><li><strong>Cost per Exception: </strong>How much does each escalation cost to resolve?</li><li><strong>Cost per Successful Task: </strong>What does the complete workflow cost after inference, retries, review, and rework?</li><li><strong>Repeat Exception Rate:</strong> How many escalations come from problems the organization has already seen before?</li></ul><p>Looking at all of these costs gives us a much better basis for understanding whether agents are actually cost-effective for that process, or whether a simpler and cheaper alternative, such as predictive AI or traditional automation, may make more economic sense.</p><h3>It’s not about taking people out of the equation</h3><p>The point of agentic AI is not to leave agents running wild with no human oversight.</p><p><a href="https://inclusioncloud.com/services/">Engineers and domain specialists</a> still have an important role to play when a case involves real uncertainty, business judgment, compliance, or higher-risk decisions.</p><p>The problem starts when their time gets absorbed by an endless exception backlog because the data, business rules, or architecture around the agent are not working as they should.</p><p>At that point, the economics can stop making sense.</p><p>The challenge is finding the right balance between what agents can handle reliably and where human supervision is actually worth the cost.</p><p>If you are evaluating an agentic AI use case, we’d be glad to learn more about the process and help you assess where it can be optimized.</p><p><a href="https://inclusioncloud.com/contact-us/">Book a discovery call</a> and tell us about your business case. We can help you evaluate the workflow, the role of HITL, and the overall economics so the process can scale without adding unnecessary cost and latency.</p><p><em>Originally published at </em><a href="https://inclusioncloud.com/insights/blog/the-exception-queue-could-be-killing-your-agent-roi/"><em>https://inclusioncloud.com</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9b47a1b9c372" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[MCP: Connecting Agents to Enterprise Systems]]></title>
            <link>https://medium.com/@inclusion.cloud/mcp-connecting-agents-to-enterprise-systems-5f79498cd93e?source=rss-aa686e488265------2</link>
            <guid isPermaLink="false">https://medium.com/p/5f79498cd93e</guid>
            <category><![CDATA[mcp-server]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[ai-gateway]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[mcp-protocol]]></category>
            <dc:creator><![CDATA[Inclusion Cloud ]]></dc:creator>
            <pubDate>Wed, 16 Sep 2026 12:56:27 GMT</pubDate>
            <atom:updated>2026-09-16T12:56:27.269Z</atom:updated>
            <content:encoded><![CDATA[<p>When Anthropic introduced the Model Context Protocol (MCP) in November 2024, the problem it was trying to solve was relatively easy to describe.</p><p>AI models were becoming increasingly capable, but they were still isolated from the systems where companies actually work. Connecting an assistant to a database, CRM, development environment, or enterprise app usually required a custom integration. Add another agent or another system, and another integration had to be built.</p><p>MCP proposed a <strong>common interface</strong> instead.</p><p><a href="https://www.anthropic.com/news/model-context-protocol">Anthropic described</a> it as an open standard for connecting AI systems with external data sources and tools. A little more than a year later, MCP had expanded far beyond the Claude ecosystem. In December 2025, Anthropic donated the protocol to the Agentic AI Foundation under the Linux Foundation, alongside support from companies including OpenAI, Google, Microsoft and AWS.</p><p>The protocol has continued to mature. The July 2026 specification moved MCP toward a stateless core designed to make servers easier to scale, route, cache and govern in production infrastructure.</p><p>MCP is increasingly showing up in enterprise AI architecture diagrams next to another rapidly growing component: the <strong>AI Gateway</strong>.</p><p>They are often discussed together. They are not the same thing.</p><p>The simplest distinction is this:</p><p><strong>MCP standardizes how AI accesses tools. An AI Gateway controls how AI traffic is routed, secured, observed and governed.</strong></p><h3>What MCP Actually Is</h3><p>One source of confusion is that “MCP” can refer to several different things in the same conversation.</p><p>At the highest level, <strong>MCP is a protocol</strong>.</p><p>It defines a standardized way for AI applications to communicate with external capabilities. The protocol establishes how clients and servers exchange messages, discover available capabilities and invoke them.</p><p>Within that architecture, three components matter most:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*1SpTcQt5NmTjXTpS.png" /></figure><p>An MCP server can expose capabilities such as<strong> tools, resources, and prompts</strong>. Tools allow the model to execute functions, while resources provide structured context or data.</p><p>Take a procurement agent.</p><p>The underlying ERP might already have dozens of APIs for inventory, suppliers, purchase requisitions and approvals. Instead of expecting the agent to understand each API independently, an MCP server could expose a smaller set of business-oriented tools:</p><ul><li>Check_inventory</li><li>Find_approved_supplier</li><li>Create_purchase_requisition</li><li>Check_purchase_order_status</li></ul><p>The agent discovers those tools through the MCP interface and invokes the appropriate one.</p><p>Importantly, <strong>MCP does not replace the underlying API.</strong></p><p>The architecture can still look like this:</p><p><strong>AI Agent → MCP Client → MCP Server → REST API → ERP</strong></p><p>MCP is the AI-facing interface. The API remains the software-facing interface underneath it.</p><p>This distinction becomes particularly valuable when several AI applications need access to the same capability. Instead of implementing the SAP inventory API separately for five different agents, the enterprise can expose the capability through one standardized MCP interface.</p><h3>So where does the AI Gateway fit?</h3><p>An AI Gateway solves a different problem.</p><p>Imagine that the same procurement agent can use GPT, Claude and Gemini. The company wants to choose which model handles each request, manage provider credentials, apply rate limits, observe latency, record usage and potentially route traffic based on cost or performance.</p><p>That is traditionally the territory of the AI Gateway.</p><p>Modern AI Gateways provide a central layer between applications and AI providers. They can normalize model access, route requests between providers, apply policies and provide centralized visibility into AI traffic.</p><p>The two layers therefore answer different architectural questions:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*PYAvd0omrsrnM2rG.png" /></figure><p>Another way to think about it:</p><p><strong>AI Gateway manages access to intelligence.</strong> <strong>MCP manages access to capabilities.</strong></p><p>A simplified architecture could look like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/936/0*0hRzwq82JSsA6mUg.png" /></figure><p>The agent sits in the middle. On one side, it needs models to reason. On the other, it needs tools to act.</p><h3>The lines are starting to blur</h3><p>The conceptual distinction remains important, but the product landscape is already becoming more complicated.</p><p>AI Gateways are expanding beyond model traffic.</p><p>Some platforms now apply the same gateway principles to MCP traffic, providing authentication, authorization, rate limiting, observability and policy enforcement in front of MCP servers.</p><p>This produces an architecture closer to:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*VhBQp52qF0wJhivn.png" /></figure><p>That does not mean AI Gateway and MCP have become the same thing.</p><p>The distinction is <strong>similar to HTTP and an API gateway</strong>.</p><p>HTTP defines how systems communicate. An API gateway governs the traffic flowing through those interfaces.</p><p>In the same way:</p><p><strong>MCP is the interoperability protocol.</strong> <strong>The gateway is the control layer around it.</strong></p><h3>MCP Is Not AI Governance</h3><p>Connecting an agent to an enterprise system through MCP does not automatically make that connection secure or well governed.</p><p>For an enterprise, protocol-level security is only the beginning.</p><p>A production environment also needs to answer questions such as:</p><ul><li>Who owns this MCP server?</li><li>Is the server approved?</li><li>Which agents can access it?</li><li>Which users can invoke a specific tool?</li><li>Should some actions require human approval?</li><li>What data can the tool return?</li><li>What happened during a particular agent execution?</li><li>Which version of the tool was used?</li></ul><p>That is AI governance around MCP, rather than MCP itself.</p><p>A mature architecture may therefore add several controls:</p><p><strong>MCP Registry:</strong> Which MCP servers and tools are approved?</p><p><strong>Gateway:</strong> Who can call them, and under what policies?</p><p><strong>Identity:</strong> On whose behalf is the agent operating?</p><p><strong>Human-in-the-loop controls:</strong> Which actions require approval?</p><p><strong>Observability:</strong> What happened during the interaction?</p><p><strong>Audit:</strong> Who initiated the action, which agent executed it and what was the outcome?</p><p>MCP makes tools interoperable. Governance determines whether they should be used.</p><h3>MCP Also Creates an AI FinOps Problem</h3><p>AI FinOps is often reduced to one question: how many tokens are we consuming?</p><p>That becomes incomplete once agents start taking actions.</p><p>An agentic workflow can generate cost across several layers:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*YX3FsilBFncOWQJX.png" /></figure><p>Consider an agent that costs only a few cents in model inference but makes 30 unnecessary tool calls, triggers an expensive data query and retries the same workflow several times.</p><p>Looking only at token consumption would miss most of the cost.</p><p>MCP therefore does not provide AI FinOps, but it creates an important new source of telemetry for it.</p><p>Enterprises increasingly need visibility not just into <strong>tokens per request</strong>, but into:</p><ul><li><strong>Tool calls per task</strong></li><li><strong>Cost per tool call</strong></li><li><strong>Retries</strong></li><li><strong>Latency</strong></li><li><strong>Failure rates</strong></li><li><strong>Model plus tool cost per completed outcome</strong></li></ul><p>That also explains why putting a gateway or observability layer around MCP traffic can become valuable. A company may eventually want to trace one business transaction from the model decision all the way through the tool execution.</p><p>The unit of economics becomes the <strong>business task</strong>, not simply the token.</p><h3>And Where Does A2A Fit?</h3><p>MCP solves another problem that is sometimes confused with agent interoperability.</p><p>Suppose a procurement agent needs help from a specialized supplier-risk agent.</p><p>The procurement agent could theoretically expose the other agent as a tool. But that starts to break down when the interaction becomes more complex.</p><p>Perhaps the supplier-risk agent needs to ask follow-up questions, perform several steps, send progress updates, negotiate what work it will perform, return multiple artifacts or maintain a longer-running interaction.</p><p>That is where <strong>Agent2Agent, or A2A, </strong>enters the architecture.</p><p>The difference is straightforward:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*gT6iv02m1vvIkIlU.png" /></figure><p>Consider a supply-chain process:</p><p><strong>Procurement Agent</strong></p><p>uses MCP → SAP procurement tools</p><p><strong>Supplier Risk Agent</strong></p><p>uses MCP → risk databases and external data</p><p><strong>Procurement Agent</strong></p><p>uses A2A → Supplier Risk Agent</p><p>The tools remain behind MCP. The agents communicate through A2A.</p><p>That separation can become increasingly important as enterprises deploy agents from different vendors.</p><h3>Is MCP Fundamental for Every Enterprise?</h3><p>Not necessarily.</p><p>The wrong conclusion from MCP’s rapid adoption would be that every API should immediately become an MCP server.</p><p>For some workloads, the existing integration is already the right architecture.</p><p>A deterministic system-to-system process does not become better simply because an AI agent is added in the middle. High-volume ETL, scheduled integrations and predictable transactional workflows will often remain better suited to APIs, event-driven architectures, integration platforms and conventional automation.</p><p>The question is not, <strong>“Can we expose this through MCP?”</strong></p><p>The better question is, <strong>“Does standardizing this capability for AI consumers create enough value to justify it?”</strong></p><p>A practical framework looks like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*55hXXERSeFKtHady.png" /></figure><p>The tipping point is usually <strong>reuse</strong>.</p><p>If one agent calls one API, MCP may simply add another layer.</p><p>If 20 agents from several platforms need access to hundreds of enterprise capabilities, standardization becomes much more valuable.</p><p>Without it, the organization risks building an increasingly complex integration matrix:</p><p><strong>many agents × many systems × many custom connectors.</strong></p><p>MCP turns that into a more reusable model:</p><p><strong>many agents → standardized tool interfaces → enterprise systems.</strong></p><h3>Ready for the Next Step in Your AI Roadmap?</h3><p>The pressure on IT leaders to move AI initiatives beyond experimentation and start showing real business results is growing.</p><p>For nearly two decades, <a href="https://www.linkedin.com/company/2409140/admin/dashboard/">Inclusion Cloud</a> has worked alongside Fortune 500 companies to turn technology initiatives into production-ready solutions. Today, our <strong>certified specialists</strong> can help you accelerate AI projects across every layer of the architecture.</p><p>Whether you are defining your architecture, moving a validated use case into production or scaling what already works, we can help you identify the right next step.</p><p><a href="https://inclusioncloud.com/contact-us/">Book a call with our team</a> to see where your AI roadmap goes next.</p><p><em>Originally published at </em><a href="https://inclusioncloud.com/insights/blog/mcp-vs-ai-gateway-enterprise-ai/"><em>https://inclusioncloud.com</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5f79498cd93e" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>