<?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 Gautam Guha on Medium]]></title>
        <description><![CDATA[Stories by Gautam Guha on Medium]]></description>
        <link>https://medium.com/@gautam2k?source=rss-82b8fabb40c7------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*xjun5cqKddN6A_Bn2CLOzQ.jpeg</url>
            <title>Stories by Gautam Guha on Medium</title>
            <link>https://medium.com/@gautam2k?source=rss-82b8fabb40c7------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 21:18:49 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@gautam2k/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[The Role of Enterprise UX Designers in the Era of Systems of Orchestration]]></title>
            <link>https://medium.com/@gautam2k/the-role-of-enterprise-ux-designers-in-the-era-of-systems-of-orchestration-f8794b0355d6?source=rss-82b8fabb40c7------2</link>
            <guid isPermaLink="false">https://medium.com/p/f8794b0355d6</guid>
            <dc:creator><![CDATA[Gautam Guha]]></dc:creator>
            <pubDate>Thu, 05 Mar 2026 11:10:57 GMT</pubDate>
            <atom:updated>2026-03-05T11:10:57.629Z</atom:updated>
            <content:encoded><![CDATA[<h3><strong>There’s a loud narrative circulating in tech circles,</strong></h3><blockquote>&quot;AI agents will do everything. There will be no screens, so there will be no need for UX Designers.&quot;</blockquote><p>Yet <a href="https://www.gartner.com/en/newsroom/press-releases/2025-06-25-gartner-predicts-over-40-percent-of-agentic-ai-projects-will-be-canceled-by-end-of-2027">Gartner</a> predicts that over 40% of agentic AI projects will be canceled by 2027 — not because the technology failed, but because of unclear business value and inadequate risk controls.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*NKljINoI1PKXDs-yFgAwsQ.png" /><figcaption>Image generated through Chat GPT</figcaption></figure><p>To my fellow Enterprise UX colleagues and leaders — The answer might be closer to home than you think.</p><p><strong>The Cost of &quot;Designer-less&quot; AI</strong><br>Over 40% of agentic AI projects will be canceled by the end of 2027, due to escalating costs, unclear business value, or inadequate risk controls, according to Gartner, Inc.<br>It isn’t because Engineers aren’t smart enough, users aren’t giving the right prompts, or Models are underperforming.<br>I propose it’s because we’re missing a critical discipline: the “UX of Intelligence.”</p><p><strong>Consider what Gartner identified as failure causes:</strong></p><p><strong>Unclear business value: </strong>Users can’t understand what the AI is doing<br><strong>Inadequate risk controls: </strong>Humans can’t intervene when AI makes mistakes<br><strong>Escalating costs: </strong>Systems aren’t designed for efficient model usage.<br>These aren’t technical problems. They’re human-AI interface problems.</p><h3>How UX Design Enables Enterprise AI</h3><p>In the paradigm of Systems of Orchestration, the UX Designer’s scope has expanded into four critical areas that determine Profit &amp; Loss (P&amp;L) success:<br><strong>1. From Interaction Design to &quot;Intent Design&quot; (Ending Technical Gatekeeping)</strong><br>Innovation is currently strangled because domain experts—planners or product owners who understand the business—must wait for IT to &quot;wire&quot; every automation.<br><strong>The UX Role:</strong> We design the Abstraction Layer. Our job is to make the plumbing (OData, JSON, API schemas) invisible. We create environments where a manager’s Domain Knowledge becomes the only &quot;code&quot; required to deploy an agent.</p><p><strong>2. From User Flow to &quot;Reasoning Governance&quot; (Breaking Rigidity)<br></strong>Traditional Enterprise Applications are deterministic and brittle. When the script fails, the system stops. The nature of work is shifting from deterministic form-filling to intent-based, probabilistic decision-making.<br><strong>The UX Role:</strong> We design Guardrails for Probabilistic Reasoning. We determine how an agent communicates a “judgment call” back to humans. We design the “Trust Interface” — the mechanism that allows humans to audit an agent’s goal-seeking logic in milliseconds, not minutes.</p><p><strong>3. From Single-Page Apps to &quot;Orchestration War Rooms&quot; (Solving Silos)</strong><br>Enterprise application like Supply chains are crippled by functional silos. Procurement doesn’t know what Finance is doing.<br><strong>The UX Role:</strong> We are architects of the Unified Worldview. We design how multi-agent systems synchronize. When an agent in Logistics acts, the UX Designer ensures the ripple effect is visible and actionable across the entire Product Structure—instantly.</p><p><strong>4. From Aesthetic to &quot;Economic Integrity&quot; (Stopping Token Leakage)<br></strong>Unsupervised agents create a &quot;Budget Black Hole&quot; through inefficient reasoning loops and unnecessary API/MCP calls.<br>The UX Role: We design for Fiscally Responsible AI. Through “Model Routing” interfaces, we help systems decide when to use a lightweight model for routine tasks versus an expensive frontier model for strategic decisions. UX has become a tool for Cloud Cost Optimization.</p><p><strong>Why This Matters Now: <br></strong>The shift from systems of record to systems of orchestration represents the most significant enterprise architecture change since the move from mainframes to client-server.</p><p>But there’s a gap, engineering teams are building sophisticated AI infrastructure while three critical design disciplines are underdeveloped. The POC failure rate isn’t about model capability — it’s about the absence of three design disciplines.</p><p><strong>Explainability Design:</strong> Users need to understand why an agent made a decision before they’ll trust it with consequential actions.<br><strong>Intervention Design:</strong> The system must gracefully handle the moment when human judgment needs to override algorithmic confidence.<br><strong>Feedback Loop Design: </strong>Agents must learn from human corrections without requiring data science expertise to retrain models.</p><p>These are inherently UX challenges. When they’re missing, we see the symptoms Gartner identified: unclear value, inadequate controls, cost overruns. No amount of prompt engineering or model fine-tuning will solve human-AI trust gaps.</p><h3><strong>The Path Forward</strong></h3><p>We are not replacing humans; we are elevating them. By stripping away the manual data-to-decision “drag,” we empower enterprise leaders to stop firefighting and start governing.</p><p>The future of Enterprise Software is not a table of parts or a series of dashboards. It is a synchronized symphony of autonomous actions.</p><p>Like a conductor leading an orchestra, the UX Designer doesn’t play every instrument — but ensures every agent plays at the right time, in the right sequence, at the right intensity. Each function (Procurement, Finance, Logistics) performs well independently, but without orchestration design, they create noise instead of symphony.</p><p><strong>The “Agentic Era” doesn’t mean the end of UX Design. It means UX has finally moved to the C-Suite.</strong></p><figure><img alt="The “Agentic Era” doesn’t mean the end of UX Design. It means UX has finally moved to the C-Suite." src="https://cdn-images-1.medium.com/max/1024/1*Dk04gAdwakVn01PtFTUJAg.png" /><figcaption>Image generated through Chat GPT</figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f8794b0355d6" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Bible of Data Visualization Studies]]></title>
            <link>https://medium.com/@gautam2k/the-bible-of-data-visualization-studies-3108b30a0b96?source=rss-82b8fabb40c7------2</link>
            <guid isPermaLink="false">https://medium.com/p/3108b30a0b96</guid>
            <dc:creator><![CDATA[Gautam Guha]]></dc:creator>
            <pubDate>Sat, 18 Jan 2025 15:22:13 GMT</pubDate>
            <atom:updated>2025-01-18T15:22:13.497Z</atom:updated>
            <content:encoded><![CDATA[<p>We are living in the world of data. The person owning maximum data, is considered to be most powerful. Visualizing and generating the right insight out of it is the job of a data analyst/ visualizer as A Picture is Worth a Thousand Words.<br>Let us go 152 years back. In 1869 Charles Joseph Minard (1781–1870) , a French civil engineer and information graphics specialist drew a piece of data visualization on Napoleon’s 1812 March. This data visualization was made fifty years after the event, when Minard was 80 years old, the map has been called “The Bible of data visualization studies”.</p><p>“Probably the best statistical graphic ever drawn.” — Edward Tufte</p><p>Let us know the history behind this data visualization.<br>This is the story of an invasion. Napoleon gathered a massive army<br>of over 400,000 to attack Russia in June 1812 from France. Napoleon’s army was more powerful, so Alexander planned to retreat as Napoleon’s troops move forward. While retreating, Alexander’s army burned everything. As a result Napoleon’s army suffered from major losses of food, disease and cold and planned to return to France.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*RqS3CaKMFFOYcPSM_aLVjg.png" /><figcaption>Nepoleon’s 1812 March by Charles Minard, 1869</figcaption></figure><p><strong>The story is depicted in the map</strong></p><blockquote>This visualization shows six types of information: geography, time, temperature, the course and direction of the army’s movement, and the number of troops remaining.</blockquote><blockquote>The widths of the gold (outward) and black (returning) paths represent the size of the force, one millimeter to 10,000 men.</blockquote><blockquote>Geographical features and major battles are marked and named, and plummeting temperatures on the return journey are shown along the bottom.</blockquote><blockquote>Napoleon’s grand army set out with a force of 400,000; only 100,000 reached Moscow; and only 10,000 returned</blockquote><p>When we start building data visualization 1st job is to identify the dimensions and measures and build various reports after defining the KPI and trying to tell a story. Sometimes multiple visualizations in a dashboard deviate from the story and people try to interpret it based on their knowledge. Here the best part is single visualization is telling me the entire story. It is 150 years old piece of precious art.</p><p><em>Source: </em><a href="https://en.wikipedia.org/wiki/Charles_Joseph_Minard"><em>https://en.wikipedia.org/wiki/Charles_Joseph_Minard</em></a><em><br>https://thoughtbot.com/blog/analyzing-minards-visualization-of-napoleons-1812-march</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3108b30a0b96" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[From Empathy to Efficiency: Design Thinking in Prompt Engineering.]]></title>
            <link>https://medium.com/@gautam2k/from-empathy-to-efficiency-design-thinking-in-prompt-engineering-80bdff33c7ae?source=rss-82b8fabb40c7------2</link>
            <guid isPermaLink="false">https://medium.com/p/80bdff33c7ae</guid>
            <dc:creator><![CDATA[Gautam Guha]]></dc:creator>
            <pubDate>Wed, 27 Nov 2024 15:44:54 GMT</pubDate>
            <atom:updated>2024-11-27T15:44:54.121Z</atom:updated>
            <content:encoded><![CDATA[<p>Most of us are now familiar with Large Language Models (LLM), such as Open AI Chat GPT, FaceBook Llama, Google Gemini, etc. We have used it for our work, experiments, and even entertainment. Providing the right input to the LLM to generate the desired outcome on 1st go is a skill. The name of this skill is prompt engineering. It is neither a Science nor an Art but a structured methodology. Educational institutes are offering courses on “Prompt Engineering”. In the not-so-distant future, we might see Prompt Engineering being offered as a regular course as well. <strong>The challenge lies in bridging the gap between human intent and machine understanding.</strong> We can greatly improve the intuitiveness and efficiency of this approach by integrating design thinking into the practice of prompt engineering.</p><p><strong>But why is this important? </strong><br>As the famous industry leader, Andrej Karpathy, former Director of AI at Tesla said, “<em>The better your prompt, the better the model will perform.</em>” This statement indicates the performance of the model is directly proportional to prompt engineering.<br> <br><strong>Why writing an accurate prompt in the first go is so important?</strong><br>Well-Designed Prompts: Better Customer Experience, Cost-Efficiency and Sustainability.<br> <br>Better Customer Experience: Improved prompt means better customer experience as it saves time. For example, asking the AI chatbot to give the right prompt to resolve the customer issue without further escalation.</p><p>Cost efficiency: Most business AI solution providers are now charging their customers based on a number of prompts they have used. More no. of prompts means a higher pricing model.</p><p>Sustainability: A substantial carbon footprint exists in running LLM and getting the desired outcome. So providing appropriate prompts is always desirable and sustainable.<br> <br><strong>Elements of a good prompt:</strong><br>We need to keep in mind the following aspects before writing a prompt.<br>1. Persona: Identify who will use the output of your prompt. For example, a financial advisor may need precise data analysis and trends from the data.<br>2. Task: Define what you want the AI to do. Like writing an email or analyzing market trends, clarity in the task is important.<br>3. Goal: What is the outcome of your prompt? <br>4. Context: Contextual information helps the AI to generate more precise outcomes. <br>5. Steps: Structured the detailed instructions into smaller steps and arrange them based on priority.<br>6. Clarity &amp; Consistency: The prompt should be less ambiguous and the language should be consistent.<br><br><strong>What is Design Thinking?</strong><br>Design thinking is a problem-solving framework based on a human-centered approach. It emphasizes empathy, ideation, prototyping, and experimentation. This helps to create efficient and model-friendly prompts.</p><p><strong>How can we correlate elements of design thinking with elements of prompt engineering?</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/735/1*5Mar7M2pwMRs8fSSd0NP0w.png" /><figcaption><strong>Correlate elements of design thinking with elements of prompt engineering.</strong></figcaption></figure><p>1. Empathy (Persona): Understand the end-users and their needs.<br>2. Define (Context): Clearly define the problem statement and context.<br>3. Ideation (Task &amp; Goal): Generate ideas and establish the task and goals.<br>4. Prototype (Steps): Design the prompt by breaking it into multiple sub-steps.<br>5. Test (Clarity &amp; Consistency): Test for clarity and consistency and reiterate if find any ambiguity.<br> <br>Applying the design thinking approach in prompt engineering helps prompt engineers to write accurate prompts which helps the model to generate better output and leads to better customer satisfaction.<br>To summarize, the following are do’s and don’ts for Prompt Engineering</p><p><strong>Do’s:<br></strong>1. Be Specific: Define the task to the point.<br>2. Use Clear Language: Avoid ambiguity.<br>3. Test Prompts: Analyze different versions to find the most effective one.<br>4. Include Context: Provide background information to guide the AI.<br><strong>Don’ts:</strong><br>1. Avoid Vagueness: Don’t interpret.<br>2. Don’t Overload: Keep prompts straightforward, avoid multiple tasks in one prompt.<br>3. Refrain from Assumed Knowledge: Don’t assume the AI will understand contexts without explanation.</p><p><strong>Example: <br></strong>Scenario: Email Generation</p><p>Normal Prompt: <br>Compose an email to Roger, our client, thanking her for the meeting and summarizing the next steps. Attach the project proposal document.</p><p>Improved Prompt: <br><em>Compose an email based on the following details. <br>Persona: You are a project manager at a marketing firm.<br>Task: Compose an email to Roger, a client you recently met with, thanking her for her time and summarizing the next steps discussed during the meeting. Attach the project proposal document and analyze the sales data from 2020 to 2022 to identify significant monthly sales trends in the European market.<br>Context: During the meeting with Roger, you discussed a new project proposal and reviewed sales data for the European market. You now need to follow up with an email summarizing the discussion and providing the necessary documents.<br>Instructions:<br>1. Compose a polite email to Roger, expressing gratitude for the meeting and summarizing the discussed next steps.<br>2. Attach the project proposal document to the email.<br>3. Analyze the sales data from 2020 to 2022, identify any significant monthly sales trends in the European market, and summarize the findings in the email.<br>4. Send the email to Roger.</em></p><p><strong>Conclusion:</strong><br>Adapting the Design Thinking framework into Prompt Engineering is a systematic way to generate improved prompts. Professionals across domains can benefit from understanding this approach to create better prompts by optimizing the full potential of LLM. With the advent of technology, communicating effectively with machines is becoming very crucial. Professionals across domains can benefit from understanding this approach, and by crafting better prompts, they can harness the full potential of LLMs. As technology continues to evolve, the ability to communicate effectively with machines will become as crucial as communicating with people.<br>By encompassing the principles of design thinking, we can create prompts that are not only model-friendly but also the outcome will be aligned with human needs, leading to more sustainable and cost-effective solutions in the AI-driven world.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=80bdff33c7ae" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>