<?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 Aniket Sanyal on Medium]]></title>
        <description><![CDATA[Stories by Aniket Sanyal on Medium]]></description>
        <link>https://medium.com/@cs_maverick?source=rss-f0dee53742d0------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*dNMSpNINma1wkA09HIgwXw.png</url>
            <title>Stories by Aniket Sanyal on Medium</title>
            <link>https://medium.com/@cs_maverick?source=rss-f0dee53742d0------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 05:52:29 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@cs_maverick/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[AI Research Made Practical: Two Short Guides to Agentic AI and Efficient AI]]></title>
            <link>https://medium.com/@cs_maverick/ai-research-made-practical-two-short-guides-to-agentic-ai-and-efficient-ai-7a7fc6348d4c?source=rss-f0dee53742d0------2</link>
            <guid isPermaLink="false">https://medium.com/p/7a7fc6348d4c</guid>
            <category><![CDATA[agentic-ai]]></category>
            <category><![CDATA[efficient-ai]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[aniket-sanyal]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Aniket Sanyal]]></dc:creator>
            <pubDate>Mon, 16 Nov 2026 06:20:06 GMT</pubDate>
            <atom:updated>2026-11-16T06:20:06.470Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Two compact technical guides for people who want to understand how modern AI systems are actually put together — not just what the latest model can do.</em></p><p>I’ve been reading applied research papers since 2012.</p><p>For a long time, I never really thought of that as something I would write publicly about. It was simply part of how I learned.</p><p>I would come across a paper, try to understand what was genuinely new about it, follow the references, look at how the architecture worked, and then compare all of that with what people were building in the real world.</p><p>Sometimes the paper was more interesting than the implementation.</p><p>Sometimes the implementation exposed limitations that were barely visible in the paper.</p><p>And sometimes an idea that looked relatively small on paper turned out to have much bigger implications once you started thinking about how it would fit into a real system.</p><p>Over the last couple of years, I started writing about these things more deliberately.</p><p>Partly because AI research has accelerated so much that keeping track of individual papers is no longer the hardest problem.</p><p>The harder problem is making sense of them together.</p><p>There is an enormous amount of good material out there — research papers, engineering blogs, benchmarks, model releases, architecture notes, framework documentation — but most of it understandably explains one piece of the puzzle at a time.</p><p>What I’ve always found more useful is trying to understand where that piece belongs.</p><p>What does it change?</p><p>What problem does it actually solve?</p><p>What new trade-offs does it introduce?</p><p>And if you were building a real system around it, what else would you need?</p><p>That is really where <strong>AI Research Made Practical</strong> came from.</p><p>I wanted to write short technical guides that sit somewhere between a research paper and a framework tutorial.</p><p>Not books that try to cover everything.</p><p>And not books that become obsolete when the next library version comes out.</p><p>Just practical guides to some of the ideas that I think are becoming important in how we design AI systems.</p><p>The first two are about <strong>agentic AI</strong> and <strong>efficient AI</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bMtzdNRidfyVKHgVlH7hgQ.png" /><figcaption>AI Research Made Practical</figcaption></figure><h3>A little context on where I’m coming from</h3><p>My own career has moved through quite a few different parts of the technology industry over the last sixteen-plus years.</p><p>I started closer to software engineering and large-scale technology delivery, and over time moved deeper into data, machine learning, AI, architecture, platforms, and eventually broader AI leadership.</p><p>Along the way, I’ve worked in very different environments: large corporate and enterprise programmes, consulting and client advisory, product teams, startup-style settings where you have to build quickly with limited resources, and research-oriented work where the questions are often much more exploratory.</p><p>Today, a large part of my work sits somewhere between <strong>AI strategy and actual delivery</strong> — building AI capabilities and platforms, thinking about enterprise adoption, working with engineering and data teams, shaping go-to-market ideas, evaluating emerging technologies, and trying to work out which research concepts are mature enough to become useful products or solutions.</p><p>I studied Artificial Intelligence at the <strong>University of Edinburgh</strong>, and I’ve also had the opportunity to work around applied research through <strong>innovation hubs in Ireland</strong>.</p><p>Those experiences gave me a deeper appreciation for the research itself, while my work across software engineering, enterprise delivery, consulting, product development, and AI leadership kept bringing me back to the practical realities of applying it. Over time, I found myself looking at the same ideas from different perspectives — understanding what made them interesting in research, what it would take to engineer them into working systems, and whether they made sense from a broader technology and business standpoint.</p><p>That combination of industry and research has probably shaped the way I read papers.</p><p>When you spend time around research, it is natural to focus on whether an approach improves a benchmark or pushes a particular capability forward.</p><p>Production introduces a different set of questions very quickly.</p><p>Can we actually build this?</p><p>Can we operate it reliably?</p><p>What will it cost?</p><p>How does it fit with the systems that already exist?</p><p>What happens when the input is messy, the model is wrong, an API fails, or a user does something nobody anticipated?</p><p>What are the security and governance implications?</p><p>And, particularly in a startup or product environment, is the extra sophistication actually worth building in the first place?</p><p>I’ve also seen the opposite problem in industry: teams can become so focused on delivery that interesting research gets dismissed simply because it does not immediately map to the current technology stack.</p><p>I’ve always enjoyed sitting somewhere between those two worlds.</p><p>I like understanding what is genuinely new in the research, but then asking a more practical question:</p><p><strong>What would have to be true for this idea to become useful in a real system?</strong></p><p>That perspective is probably the strongest common thread behind these books.</p><p>It is also why <strong>AI Research Made Practical</strong> felt like the right name for the series.</p><h3>The Beginner’s Field Guide to Agentic AI</h3><p>Agentic AI is one of those areas where the terminology has become broader faster than the underlying concepts have become clearer.</p><p>Depending on who you talk to, an “agent” might be an LLM with access to a tool, a workflow with a loop, a planning system, or a much larger architecture involving memory, retrieval, multiple agents, verification, security boundaries, and human approvals.</p><p>I wanted to strip some of that ambiguity away.</p><p><strong>The Beginner’s Field Guide to Agentic AI</strong> is my attempt to explain what actually sits around the model when we start asking an AI system to do more than generate an answer.</p><p>Because once a model starts taking actions, things get interesting very quickly.</p><p>It needs to know what tools it can use.</p><p>It may need to remember information from earlier steps.</p><p>It may need to retrieve external knowledge.</p><p>It needs some way of understanding whether an action succeeded.</p><p>Sometimes several agents need to coordinate.</p><p>Sometimes an action should happen automatically.</p><p>Sometimes it absolutely should not.</p><p>The book looks at planning, tool use, memory, protocols such as <strong>MCP, A2A, ACP and ANP</strong>, retrieval-augmented generation, KG-RAG, coding and data agents, multi-agent systems, security, verification, observability, approval gates, and governed autonomy.</p><p>The main idea behind the book is fairly simple:</p><p><strong>An AI agent is not just a model. It is a system.</strong></p><p>Once you start looking at it that way, a lot of the design choices make more sense.</p><h3>Small Models, Big Impact</h3><p>The second book came from a different question.</p><p>For obvious reasons, a lot of AI discussion gravitates toward bigger models.</p><p>More parameters. More capability. Bigger context windows. Bigger infrastructure.</p><p>But that is not always the direction a real product needs to go.</p><p>In practice, you also care about latency, memory, inference cost, energy usage, privacy, connectivity, and the hardware available at the point where the model needs to run.</p><p>That leads to a much less glamorous but very useful question:</p><p><strong>What is the smallest model — or the simplest overall system — that can do the job reliably?</strong></p><p>That is what <strong>Small Models, Big Impact</strong> explores.</p><p>The book covers small language models, <strong>Direct Preference Knowledge Distillation (DPKD)</strong>, quantization, parameter-efficient adaptation, TinyLoRA, TurboQuant, FlashAttention, LiteRT, inference optimization, and on-device AI.</p><p>I didn’t want the book to become an argument that small models are always better.</p><p>They aren’t.</p><p>There are plenty of problems where a large model is exactly what you want.</p><p>But there are also problems where sending everything to the largest available model is expensive, slow, unnecessary, or simply impractical.</p><p>The interesting part is figuring out where that boundary lies.</p><p>Sometimes the answer is a large model in the cloud.</p><p>Sometimes it is a smaller model running locally.</p><p>Sometimes it is routing between several models.</p><p>Sometimes <strong>routing, retrieval, adaptation, or the surrounding architecture matters more than simply moving to a larger model</strong>.</p><p>Sometimes the biggest improvement comes from changing the surrounding architecture rather than changing the model at all.</p><p>That is why I tend to think of efficient AI as a system-design problem rather than just a model-compression problem.</p><h3>Why these two books together?</h3><p>When I started working on them, I didn’t initially think of agentic AI and efficient AI as a pair.</p><p>But the more I worked through the material, the more connected they started to feel.</p><p>On one side, AI systems are becoming more capable because we are adding planning, memory, retrieval, tools, orchestration, and increasingly sophisticated control around models.</p><p>On the other side, we want those same systems to run faster, cost less, use less memory, work on more devices, and sometimes operate closer to where the data actually lives.</p><p>Those pressures eventually meet.</p><p>A very capable agent that is too expensive or too slow to use is not a particularly good system.</p><p>A tiny model that runs beautifully on a device but cannot reliably complete the task is not a particularly good system either.</p><p>The architecture has to balance both.</p><p>That is the connection I wanted the first two books in the series to capture.</p><h3>Why keep them short?</h3><p>This was deliberate.</p><p>There are already excellent textbooks, long-form technical books, papers, courses, and documentation covering many of these subjects in much greater depth.</p><p>I didn’t want to reproduce those.</p><p>I wanted something you could realistically read from beginning to end without turning it into a three-month project.</p><p>Something that gives you enough context to understand the landscape, and then makes it easier to go deeper into the papers, tools, and implementations that matter to you.</p><p>I also deliberately avoided making either book dependent on one particular framework.</p><p>Frameworks are useful, but they change quickly.</p><p>The underlying ideas usually have a longer shelf life.</p><p>Memory, retrieval, verification, tool boundaries, quantization, distillation, latency, resource constraints — these questions will still matter even when the names of the libraries change.</p><h3>Who I had in mind while writing</h3><p>I had a fairly broad group of readers in mind, partly because I’ve worked with these ideas from several sides myself — as an engineer, around research, in product and consulting environments, with enterprise teams, in startup-style settings, and now from an AI leadership perspective.</p><p>So I didn’t want the books to assume that the reader is an ML researcher.</p><p>I had software engineers in mind. Architects. Product and technology leaders. Data and ML practitioners. Consultants. Students. And people moving into AI from adjacent areas who want something more substantial than a high-level explainer without starting with twenty research papers.</p><p>If you already work deeply in these areas, some of the material will naturally be familiar.</p><p>But I hope the way the pieces are connected is still useful.</p><p>Sometimes a good mental model is valuable even when you already know the individual components.</p><h3>Where to get the books</h3><p>Both books are now available across <strong>Amazon Kindle, Google Play Books, Apple Books, and Gumroad</strong>, so you can choose whichever reading ecosystem works best for you.</p><h3>The Beginner’s Field Guide to Agentic AI</h3><p><a href="https://www.amazon.com/dp/B0HKLFFYYD"><strong>Amazon Kindle</strong></a> · <a href="https://play.google.com/store/books/details?id=L8kOEgAAQBAJ"><strong>Google Play Books</strong></a> · <a href="https://books.apple.com/us/book/the-beginners-field-guide-to-agentic-ai/id6814144869"><strong>Apple Books</strong></a> · <a href="https://csmaverick.gumroad.com/l/agentic_ai"><strong>Gumroad</strong></a></p><h3>Small Models, Big Impact</h3><p><a href="https://www.amazon.com/dp/B0HKLFMQ4J"><strong>Amazon Kindle</strong></a> · <a href="https://play.google.com/store/books/details?id=fMkOEgAAQBAJ"><strong>Google Play Books</strong></a> · <a href="https://books.apple.com/us/book/small-models-big-impact/id6814147386"><strong>Apple Books</strong></a> · <a href="https://csmaverick.gumroad.com/l/efficient_ai"><strong>Gumroad</strong></a></p><h3>Get both books together</h3><p>If you’re interested in both topics, I’ve also put the two books together as a digital bundle.</p><p><strong>AI Research Made Practical — Complete 2-Book Digital Bundle</strong></p><p>It includes both books in <strong>PDF and EPUB formats</strong>, so you can read them in whichever format works best for you.</p><p><a href="https://csmaverick.gumroad.com/l/ai_bundle"><strong>Get the complete two-book bundle on Gumroad</strong></a></p><h3>And this is just the start</h3><p>There are quite a few areas I want to explore next.</p><p>Retrieval systems. Context engineering. Evaluation. AI security. Memory architectures. Model routing. Multimodal systems. AI infrastructure. Production patterns for systems that are becoming increasingly autonomous.</p><p>A lot of these areas have the same problem: there is plenty of information available, but it takes time to connect the research to the architecture and the architecture to something you could actually build.</p><p>That is the part I enjoy.</p><p>After all these years of reading research, I still find myself coming back to the same questions:</p><p><strong>What is the core idea here?</strong></p><p><strong>Where does it fit?</strong></p><p><strong>What does it improve?</strong></p><p><strong>What does it make harder?</strong></p><p><strong>And what would it take to make it genuinely useful outside the paper?</strong></p><p>That is what I want <strong>AI Research Made Practical</strong> to be about.</p><p>These two books are where I decided to start.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7a7fc6348d4c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[CUDA Agent: The AI Agent That Learned to Outrun torch.compile at GPU Kernel Optimization]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-snippet">How agentic reinforcement learning is turning LLMs from code generators into hardware-aware GPU performance engineers</p><p class="medium-feed-link"><a href="https://ai.plainenglish.io/cuda-agent-the-ai-agent-that-learned-to-outrun-torch-compile-at-gpu-kernel-optimization-771f195349dd?source=rss-f0dee53742d0------2">Continue reading on Artificial Intelligence in Plain English »</a></p></div>]]></description>
            <link>https://ai.plainenglish.io/cuda-agent-the-ai-agent-that-learned-to-outrun-torch-compile-at-gpu-kernel-optimization-771f195349dd?source=rss-f0dee53742d0------2</link>
            <guid isPermaLink="false">https://medium.com/p/771f195349dd</guid>
            <category><![CDATA[gpu-optimization]]></category>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[reinforcement-learning]]></category>
            <category><![CDATA[cuda]]></category>
            <dc:creator><![CDATA[Aniket Sanyal]]></dc:creator>
            <pubDate>Sun, 13 Sep 2026 07:06:06 GMT</pubDate>
            <atom:updated>2026-09-21T16:14:00.843Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Beyond RPA: How Agentic AI Is Rewiring Enterprise Accounting, Finance Automation, and the Future of…]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-snippet">From software bots that follow instructions to AI agents that pursue goals, enterprise automation is entering a very different era.</p><p class="medium-feed-link"><a href="https://medium.datadriveninvestor.com/beyond-rpa-how-agentic-ai-is-rewiring-enterprise-accounting-finance-automation-and-the-future-of-d47280313c05?source=rss-f0dee53742d0------2">Continue reading on DataDrivenInvestor »</a></p></div>]]></description>
            <link>https://medium.datadriveninvestor.com/beyond-rpa-how-agentic-ai-is-rewiring-enterprise-accounting-finance-automation-and-the-future-of-d47280313c05?source=rss-f0dee53742d0------2</link>
            <guid isPermaLink="false">https://medium.com/p/d47280313c05</guid>
            <category><![CDATA[rap]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[enterprise-ai]]></category>
            <category><![CDATA[agentic-ai]]></category>
            <dc:creator><![CDATA[Aniket Sanyal]]></dc:creator>
            <pubDate>Sun, 13 Sep 2026 07:04:59 GMT</pubDate>
            <atom:updated>2026-09-29T11:59:33.559Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[AgentGuardian: The AI Agent Security Layer That Learns What Your Agent Is Allowed to Do]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-snippet">Why the next frontier of AI security may be controlling tool calls, permissions, and execution paths &#x2014; not just filtering prompts</p><p class="medium-feed-link"><a href="https://ai.plainenglish.io/agentguardian-the-ai-agent-security-layer-that-learns-what-your-agent-is-allowed-to-do-2f5d7eaf9310?source=rss-f0dee53742d0------2">Continue reading on Artificial Intelligence in Plain English »</a></p></div>]]></description>
            <link>https://ai.plainenglish.io/agentguardian-the-ai-agent-security-layer-that-learns-what-your-agent-is-allowed-to-do-2f5d7eaf9310?source=rss-f0dee53742d0------2</link>
            <guid isPermaLink="false">https://medium.com/p/2f5d7eaf9310</guid>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[ai-governance]]></category>
            <category><![CDATA[ai-security]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Aniket Sanyal]]></dc:creator>
            <pubDate>Sat, 05 Sep 2026 14:50:28 GMT</pubDate>
            <atom:updated>2026-09-05T14:50:28.668Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Agent-X Exposes the Reality of Agentic AI: Why Multimodal AI Agents Still Struggle With Vision…]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-snippet">AI agents can see images, watch videos, search the web, call tools, solve equations, and generate answers. Agent-X asks a harder question&#x2026;</p><p class="medium-feed-link"><a href="https://medium.datadriveninvestor.com/agent-x-exposes-the-reality-of-agentic-ai-why-multimodal-ai-agents-still-struggle-with-vision-796c3109612c?source=rss-f0dee53742d0------2">Continue reading on DataDrivenInvestor »</a></p></div>]]></description>
            <link>https://medium.datadriveninvestor.com/agent-x-exposes-the-reality-of-agentic-ai-why-multimodal-ai-agents-still-struggle-with-vision-796c3109612c?source=rss-f0dee53742d0------2</link>
            <guid isPermaLink="false">https://medium.com/p/796c3109612c</guid>
            <category><![CDATA[multimodal-ai]]></category>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[agentic-ai]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[computer-vision]]></category>
            <dc:creator><![CDATA[Aniket Sanyal]]></dc:creator>
            <pubDate>Sat, 05 Sep 2026 14:50:05 GMT</pubDate>
            <atom:updated>2026-09-15T00:53:51.239Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Every AI Prompt Has a Power Bill: The Hidden Energy, Water and Carbon Cost of Generative AI]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.datadriveninvestor.com/every-ai-prompt-has-a-power-bill-the-hidden-energy-water-and-carbon-cost-of-generative-ai-2b4cab3659ca?source=rss-f0dee53742d0------2"><img src="https://cdn-images-1.medium.com/max/2600/0*BR1B9D2oz9cOZ5sN" width="5120"></a></p><p class="medium-feed-snippet">Why AI inference may matter more than training, why efficiency alone won&#x2019;t solve the problem, and how the CLEAR framework could help build&#x2026;</p><p class="medium-feed-link"><a href="https://medium.datadriveninvestor.com/every-ai-prompt-has-a-power-bill-the-hidden-energy-water-and-carbon-cost-of-generative-ai-2b4cab3659ca?source=rss-f0dee53742d0------2">Continue reading on DataDrivenInvestor »</a></p></div>]]></description>
            <link>https://medium.datadriveninvestor.com/every-ai-prompt-has-a-power-bill-the-hidden-energy-water-and-carbon-cost-of-generative-ai-2b4cab3659ca?source=rss-f0dee53742d0------2</link>
            <guid isPermaLink="false">https://medium.com/p/2b4cab3659ca</guid>
            <category><![CDATA[genai]]></category>
            <category><![CDATA[ai-infrastructure]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[sustainable-ai]]></category>
            <category><![CDATA[green-ai]]></category>
            <dc:creator><![CDATA[Aniket Sanyal]]></dc:creator>
            <pubDate>Sat, 29 Aug 2026 07:51:58 GMT</pubDate>
            <atom:updated>2026-09-01T12:08:51.490Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Debugging Agentic AI: How DiLLS Turns Multi-Agent LLM Chaos Into an Explainable System]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-snippet">A new approach to AI agent debugging replaces endless logs with three layers of insight: activity, action, and operation.</p><p class="medium-feed-link"><a href="https://ai.plainenglish.io/debugging-agentic-ai-how-dills-turns-multi-agent-llm-chaos-into-an-explainable-system-04b6c3fb2719?source=rss-f0dee53742d0------2">Continue reading on Artificial Intelligence in Plain English »</a></p></div>]]></description>
            <link>https://ai.plainenglish.io/debugging-agentic-ai-how-dills-turns-multi-agent-llm-chaos-into-an-explainable-system-04b6c3fb2719?source=rss-f0dee53742d0------2</link>
            <guid isPermaLink="false">https://medium.com/p/04b6c3fb2719</guid>
            <category><![CDATA[ai-debugging]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[ai-observability]]></category>
            <category><![CDATA[explainable-ai]]></category>
            <category><![CDATA[ai-agent]]></category>
            <dc:creator><![CDATA[Aniket Sanyal]]></dc:creator>
            <pubDate>Sat, 29 Aug 2026 07:51:16 GMT</pubDate>
            <atom:updated>2026-08-29T07:51:16.132Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Autonomous Financial DevOps: How Agentic AI Could Reinvent FinOps, Cloud Cost Optimization and…]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-snippet">From dashboards that explain overspending to AI agents that detect, decide and act &#x2014; inside financial, risk and compliance guardrails.</p><p class="medium-feed-link"><a href="https://ai.plainenglish.io/autonomous-financial-devops-how-agentic-ai-could-reinvent-finops-cloud-cost-optimization-and-7b01dda1d3a2?source=rss-f0dee53742d0------2">Continue reading on Artificial Intelligence in Plain English »</a></p></div>]]></description>
            <link>https://ai.plainenglish.io/autonomous-financial-devops-how-agentic-ai-could-reinvent-finops-cloud-cost-optimization-and-7b01dda1d3a2?source=rss-f0dee53742d0------2</link>
            <guid isPermaLink="false">https://medium.com/p/7b01dda1d3a2</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[enterprise-ai]]></category>
            <category><![CDATA[finops]]></category>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[responsible-ai]]></category>
            <dc:creator><![CDATA[Aniket Sanyal]]></dc:creator>
            <pubDate>Sat, 22 Aug 2026 15:22:47 GMT</pubDate>
            <atom:updated>2026-08-22T15:22:47.429Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[The Agentic Enterprise Playbook: How to Scale AI Agents Without Losing Human Control]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-snippet">What SCORE-AI, the L1&#x2013;L6 Agent Maturity Model, AURA, and RRR can teach us about AI governance, autonomous agents, and the future of&#x2026;</p><p class="medium-feed-link"><a href="https://medium.datadriveninvestor.com/the-agentic-enterprise-playbook-how-to-scale-ai-agents-without-losing-human-control-f40d333ce62f?source=rss-f0dee53742d0------2">Continue reading on DataDrivenInvestor »</a></p></div>]]></description>
            <link>https://medium.datadriveninvestor.com/the-agentic-enterprise-playbook-how-to-scale-ai-agents-without-losing-human-control-f40d333ce62f?source=rss-f0dee53742d0------2</link>
            <guid isPermaLink="false">https://medium.com/p/f40d333ce62f</guid>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[ai-governance]]></category>
            <category><![CDATA[responsible-ai]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[enterprise-ai]]></category>
            <dc:creator><![CDATA[Aniket Sanyal]]></dc:creator>
            <pubDate>Sat, 22 Aug 2026 15:18:02 GMT</pubDate>
            <atom:updated>2026-08-25T05:06:44.859Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Why AI Agent Teams Get Stuck]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://pub.towardsai.net/why-ai-agent-teams-get-stuck-ec94750bd995?source=rss-f0dee53742d0------2"><img src="https://cdn-images-1.medium.com/max/915/1*JmE6hCgvjSipYRw_f3wt9g.png" width="915"></a></p><p class="medium-feed-snippet">Multi-Agent LLMs, MACE, and the Hidden Collaboration Problem</p><p class="medium-feed-link"><a href="https://pub.towardsai.net/why-ai-agent-teams-get-stuck-ec94750bd995?source=rss-f0dee53742d0------2">Continue reading on Towards AI »</a></p></div>]]></description>
            <link>https://pub.towardsai.net/why-ai-agent-teams-get-stuck-ec94750bd995?source=rss-f0dee53742d0------2</link>
            <guid isPermaLink="false">https://medium.com/p/ec94750bd995</guid>
            <category><![CDATA[ai-orchestration]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[multi-agent-systems]]></category>
            <category><![CDATA[ai-research]]></category>
            <dc:creator><![CDATA[Aniket Sanyal]]></dc:creator>
            <pubDate>Sun, 02 Aug 2026 12:30:09 GMT</pubDate>
            <atom:updated>2026-08-02T16:06:02.820Z</atom:updated>
        </item>
    </channel>
</rss>