<?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 Dennis Layton on Medium]]></title>
        <description><![CDATA[Stories by Dennis Layton on Medium]]></description>
        <link>https://medium.com/@dlaytonj2?source=rss-fce655d290ee------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*PDF5-2ePkwXFJBdz.</url>
            <title>Stories by Dennis Layton on Medium</title>
            <link>https://medium.com/@dlaytonj2?source=rss-fce655d290ee------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 16:31:22 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@dlaytonj2/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[Enterprise AI Adoption: The AI Vendor Narrative of Capability Overhang]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@dlaytonj2/enterprise-ai-adoption-the-ai-vendor-narrative-of-capability-overhang-8af092d90579?source=rss-fce655d290ee------2"><img src="https://cdn-images-1.medium.com/max/1280/0*8mXn3EBNyzf3Ecn3" width="1280"></a></p><p class="medium-feed-snippet">The AI model wars are unimportant. The harder problem &#x2014; the one nobody&#x2019;s solved yet &#x2014; is making AI adoption actually pay off at enterprise&#x2026;</p><p class="medium-feed-link"><a href="https://medium.com/@dlaytonj2/enterprise-ai-adoption-the-ai-vendor-narrative-of-capability-overhang-8af092d90579?source=rss-fce655d290ee------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@dlaytonj2/enterprise-ai-adoption-the-ai-vendor-narrative-of-capability-overhang-8af092d90579?source=rss-fce655d290ee------2</link>
            <guid isPermaLink="false">https://medium.com/p/8af092d90579</guid>
            <category><![CDATA[data]]></category>
            <category><![CDATA[claude]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[chatgpt]]></category>
            <dc:creator><![CDATA[Dennis Layton]]></dc:creator>
            <pubDate>Thu, 30 Apr 2026 19:04:07 GMT</pubDate>
            <atom:updated>2026-04-30T19:04:07.374Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Does Your AI Coding Agent Need to Sleep?]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@dlaytonj2/does-your-ai-coding-agent-need-to-sleep-b56ea751c442?source=rss-fce655d290ee------2"><img src="https://cdn-images-1.medium.com/max/1232/1*wKY486ADLn8ilGAOCavj9Q.png" width="1232"></a></p><p class="medium-feed-snippet">Claude Code&#x2019;s quietly shipped memory consolidation feature, Autodream, and what it reveals about the limits of persistent AI context.</p><p class="medium-feed-link"><a href="https://medium.com/@dlaytonj2/does-your-ai-coding-agent-need-to-sleep-b56ea751c442?source=rss-fce655d290ee------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@dlaytonj2/does-your-ai-coding-agent-need-to-sleep-b56ea751c442?source=rss-fce655d290ee------2</link>
            <guid isPermaLink="false">https://medium.com/p/b56ea751c442</guid>
            <category><![CDATA[openai]]></category>
            <category><![CDATA[data]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[claude-code]]></category>
            <category><![CDATA[claude]]></category>
            <dc:creator><![CDATA[Dennis Layton]]></dc:creator>
            <pubDate>Thu, 02 Apr 2026 17:56:40 GMT</pubDate>
            <atom:updated>2026-04-02T18:00:47.804Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Holding Two Truths at Once: AI, Hype, and a Breakthrough in a Lab]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@dlaytonj2/holding-two-truths-at-once-ai-hype-and-a-breakthrough-in-a-lab-c0f37f8d8f0f?source=rss-fce655d290ee------2"><img src="https://cdn-images-1.medium.com/max/1536/1*CTgMW3LhJLRi76P50Ylo2Q.jpeg" width="1536"></a></p><p class="medium-feed-snippet">We live in interesting times.</p><p class="medium-feed-link"><a href="https://medium.com/@dlaytonj2/holding-two-truths-at-once-ai-hype-and-a-breakthrough-in-a-lab-c0f37f8d8f0f?source=rss-fce655d290ee------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@dlaytonj2/holding-two-truths-at-once-ai-hype-and-a-breakthrough-in-a-lab-c0f37f8d8f0f?source=rss-fce655d290ee------2</link>
            <guid isPermaLink="false">https://medium.com/p/c0f37f8d8f0f</guid>
            <category><![CDATA[data]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[chatgpt]]></category>
            <category><![CDATA[openai]]></category>
            <dc:creator><![CDATA[Dennis Layton]]></dc:creator>
            <pubDate>Wed, 25 Mar 2026 18:27:15 GMT</pubDate>
            <atom:updated>2026-03-25T18:27:15.485Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Parallel Development with Codex: Unlocking Git Worktrees]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@dlaytonj2/parallel-development-with-codex-unlocking-git-worktrees-f9ff0401f27a?source=rss-fce655d290ee------2"><img src="https://cdn-images-1.medium.com/max/1232/0*aDqJ5nJCUMloR5Om" width="1232"></a></p><p class="medium-feed-snippet">Introduction</p><p class="medium-feed-link"><a href="https://medium.com/@dlaytonj2/parallel-development-with-codex-unlocking-git-worktrees-f9ff0401f27a?source=rss-fce655d290ee------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@dlaytonj2/parallel-development-with-codex-unlocking-git-worktrees-f9ff0401f27a?source=rss-fce655d290ee------2</link>
            <guid isPermaLink="false">https://medium.com/p/f9ff0401f27a</guid>
            <category><![CDATA[chatgpt]]></category>
            <category><![CDATA[openai]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[data]]></category>
            <dc:creator><![CDATA[Dennis Layton]]></dc:creator>
            <pubDate>Sat, 28 Feb 2026 20:00:26 GMT</pubDate>
            <atom:updated>2026-02-28T20:51:40.147Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[The Year the Lights Went Out in Software]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@dlaytonj2/the-year-the-lights-went-out-in-software-8b1ba337f8f0?source=rss-fce655d290ee------2"><img src="https://cdn-images-1.medium.com/max/1232/1*5SeAZH_3EriC74uBx_97wQ.png" width="1232"></a></p><p class="medium-feed-snippet">Before You Read this Article</p><p class="medium-feed-link"><a href="https://medium.com/@dlaytonj2/the-year-the-lights-went-out-in-software-8b1ba337f8f0?source=rss-fce655d290ee------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@dlaytonj2/the-year-the-lights-went-out-in-software-8b1ba337f8f0?source=rss-fce655d290ee------2</link>
            <guid isPermaLink="false">https://medium.com/p/8b1ba337f8f0</guid>
            <category><![CDATA[claude]]></category>
            <category><![CDATA[openai]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[chatgpt]]></category>
            <category><![CDATA[data]]></category>
            <dc:creator><![CDATA[Dennis Layton]]></dc:creator>
            <pubDate>Mon, 23 Feb 2026 00:01:25 GMT</pubDate>
            <atom:updated>2026-02-28T20:52:19.867Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Codex 5.3 vs Opus 4.6]]></title>
            <link>https://medium.com/@dlaytonj2/codex-5-3-vs-opus-4-6-61cea8eedf50?source=rss-fce655d290ee------2</link>
            <guid isPermaLink="false">https://medium.com/p/61cea8eedf50</guid>
            <category><![CDATA[chatgpt]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[openai]]></category>
            <category><![CDATA[data]]></category>
            <dc:creator><![CDATA[Dennis Layton]]></dc:creator>
            <pubDate>Thu, 12 Feb 2026 01:32:30 GMT</pubDate>
            <atom:updated>2026-02-12T01:32:30.504Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*i2mrDAM27jzkxwsXnZl1tA.png" /></figure><p>I couldn’t resist. When Codex 5.3 and Opus 4.6 dropped on the same day, I gave both coding models the exact same challenge: build a flight simulator. To make it more interesting, I set a tougher constraint — simulate the aircraft Orville and Wilbur Wright flew at Kitty Hawk.</p><p>I used ChatGPT (GPT5.2) to come up with the requirements. Here is the prompt I used for that.</p><blockquote><em>Create a prompt for an AI coding platform like Codex to generate an historically accurate simulation of the first manned flight by the Wright brothers. The simulation should allow the user to work with the same variables such as wind speed that the Wright brothers would have had to contend with. The simulator should also recreate the terrain at Kitty Hawk.</em></blockquote><p>The requirements generated are shown below. A few months ago, I would’ve then broken this into smaller chunks instead of attempting it in one shot. But today, it’s not a problem. The requirements generated by GPT-5.2 are shown below. What stood out to me is how much research has to happen before a single line of code can be written — and how naturally the model accounted for that upfront.</p><p>I was also impressed by the suggestion to support multiple camera angles. The “camera modes” section in particular felt like a real game-development touch, not something you’d expect from a first-pass requirements draft.</p><pre>You are an expert simulation engineer AND a careful historian. <br>Build an educational, historically accurate interactive simulator of the <br>Wright brothers’ FIRST successful powered, controlled, manned flight <br>(Wright Flyer I), flown at 10:35 AM on December 17, 1903, at <br>Kill Devil Hills near Kitty Hawk, North Carolina.<br><br>PRIMARY GOAL<br>Create a playable simulation where a user can attempt the first flight under historically grounded conditions, while being able to vary the same kinds of variables the Wrights contended with—especially wind speed/direction/gusts—and observe whether the aircraft lifts off, how it handles, and how far it goes.<br><br>PLATFORM &amp; TECH CHOICE (pick ONE and implement fully)<br>Option A (preferred): Web-based simulator using TypeScript + Vite + React + Three.js for rendering. Use a lightweight physics approach (custom rigid-body integrator) and implement aerodynamics yourself.<br>Option B: Python desktop app using Panda3D or PyGame + OpenGL (only if you cannot do web well).<br><br>HARD REQUIREMENTS (must have)<br>1) Historically grounded aircraft model (Wright Flyer I / “1903 Flyer”)<br>   - Geometry and mass properties: wingspan, wing area, length, empty weight, pilot prone position.<br>   - Propulsion: ~12 hp engine driving twin pusher propellers via chain drive (simplify drivetrain but reflect limited power).<br>   - Controls: canard elevator (pitch), wing-warping (roll), and coupled rudder/yaw behavior (coordinated turns were rudimentary; initial flights were essentially straight).<br>   - Ground handling: no wheels on the aircraft; it rests on skids and uses a launch dolly/truck on a 60-foot wooden launch rail/monorail. Takeoff occurs from the rail into a headwind.<br>   - Include a “historical preset” that approximates the first successful flight outcome (about 12 seconds and ~120 feet ground distance) when set to historically plausible conditions and gentle control inputs.<br><br>2) Wind + atmosphere model with user-adjustable variables<br>   - Wind speed (mph and m/s), direction, gustiness, turbulence intensity.<br>   - Preset: strong headwind (~27 mph) typical of the day.<br>   - Optional: temperature/air density slider (simple ISA-based approximation is fine).<br><br>3) Terrain recreation of Kitty Hawk / Kill Devil Hills area<br>   - Create a 3D terrain surface that plausibly matches the sandy, dune-dominated setting.<br>   - Minimum acceptable: heightmap-based dunes + sand textures + sparse vegetation.<br>   - Preferred: support loading a real DEM/heightmap (GeoTIFF or converted PNG) and/or GIS layers; provide a small “tools/” script to convert common DEM formats into a heightmap usable by the app.<br>   - Place key scene elements: launch rail on relatively level ground; nearby dune/hill presence to contextualize Kill Devil Hills; optional markers for takeoff and landing points.<br><br>4) User interface<br>   - Camera modes: chase cam, fixed tripod cam (like the famous photo angle), free-fly, top-down map view.<br>   - Controls: on-screen sliders/inputs for wind and aircraft parameters; simple keyboard/gamepad mapping for flight controls.<br>   - Instrumentation overlay (HUD): airspeed (airspeed and groundspeed), altitude, pitch/roll, time aloft, distance traveled over ground.<br>   - “Reset to Historical Conditions” button.<br>   - “Replay” feature: record control inputs and wind seed so a run can be replayed deterministically.<br><br>5) Educational “historical notes”<br>   - Include a short in-app panel or README section that lists the key historical parameters you used (aircraft specs, launch rail length, first flight distance/time, wind conditions), and note what you simplified.<br><br>PHYSICS + AERO (implement, but keep it reasonable)<br>- Use a rigid-body state: position, velocity, orientation (quaternion), angular velocity.<br>- Compute forces each timestep:<br>  - Gravity<br>  - Aerodynamic lift/drag for:<br>    - main wings (biplane treated as a combined lifting surface is acceptable)<br>    - canard elevator surface<br>    - rudder/vertical tails<br>  - Propeller thrust (simple thrust curve vs airspeed and throttle is fine)<br>  - Ground/rail contact forces while on dolly/rail (friction + constraint along rail until liftoff)<br>- Aerodynamic model:<br>  - Use Lift = 0.5*rho*V^2*S*CL(alpha), Drag = 0.5*rho*V^2*S*CD(alpha)<br>  - Provide plausible CL/CD curves; you may approximate with piecewise linear curves and a stall angle.<br>  - Include a simple coupling so wing-warping produces differential lift/roll and can induce adverse yaw (rudder helps).<br>- Numerical integration:<br>  - Fixed timestep (e.g., 1/120s) with semi-implicit Euler or RK2.<br>  - Deterministic mode when a “seed” is set for wind gusts.<br><br>HISTORICAL PRESETS (include these as scenario buttons)<br>- “First Flight (Dec 17, 1903 10:35)”:<br>  - strong headwind preset, short straight flight target, low altitude.<br>- “Four Flights of the Day”:<br>  - include presets for the approximate distances/times of the 4 flights; show as challenge modes.<br>- “Sandbox”:<br>  - user can tweak wind, mass, engine power, gusts, rail friction, etc.<br><br>ACCEPTANCE CRITERIA<br>- App runs locally with a single command (npm install; npm run dev) for web version.<br>- The historical preset produces a short hop consistent with the real event (order-of-magnitude correct: ~10–15 seconds, ~100–200 feet ground distance) without absurd control inputs.<br>- Wind changes meaningfully affect takeoff success and controllability.<br>- Terrain is clearly dune/sand terrain and not a flat plane; launch rail is present and functional.<br><br>DELIVERABLES<br>- Full source code project with clear structure<br>- README with:<br>  - how to run<br>  - what was modeled vs simplified<br>  - where to drop a DEM/heightmap file if the user wants higher fidelity<br>- Basic automated tests for:<br>  - deterministic replay<br>  - physics step stability (no NaNs; bounded orientation)<br>  - scenario presets load correctly<br><br>IMPLEMENTATION NOTES<br>- Use SI units internally (m, s, kg, N) but display mph/ft as well.<br>- No proprietary assets. Use simple procedural textures or freely usable placeholders.<br>- Keep performance smooth (target 60 FPS in browser).<br><br>Now generate the complete project (all files), with code, and a concise README.Both models delivered a working simulation almost out of the box, with only minor issues that were quickly fixed.</pre><p>Aesthetically, Opus 4.6 produced a better visual experience — the aircraft and the terrain both looked more polished. Functionally, though, Codex 5.3 performed better overall and was easier to learn and work with.</p><p>The bigger difference was in how they <em>iterated</em>. Opus 4.5 (in my testing) seemed to burn through tokens quickly. When I tried to add a throttle visualization, it took several attempts to get it right, and each attempt consumed more tokens.</p><p>Codex 5.3, by contrast, spent more time testing its changes. It could even invoke Playwright to actually run the game, verify whether a feature worked, and then fix it when it didn’t — which made the whole loop feel more efficient and reliable.</p><h3>Final Word</h3><p>There’s no clear winner here — I simply preferred working with Codex 5.3.</p><p>Because I have a Plus membership, the cost of developing the simulation was effectively absorbed into the $20 USD/month subscription. Running Claude Code with Opus 4.5, on the other hand, cost me a little over $5 USD to generate the same simulation.</p><p>To be fair, I’d expect that with a comparable Anthropic subscription, the cost experience would likely look similar as well.</p><p>The key takeaway is that both models were able to produce a simulator of this complexity in under an hour — and then work with me tirelessly to fine-tune the controls and handling, without losing context along the way.</p><p>In the process, I gained a much better appreciation for how difficult it must have been to fly that first aircraft. I’ve only managed to get it airborne a handful of times.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=61cea8eedf50" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Claude Code Tasks: When Claude Code Plans the Work (and Then Does It)]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@dlaytonj2/claude-code-tasks-when-claude-code-plans-the-work-and-then-does-it-43f98f5113cd?source=rss-fce655d290ee------2"><img src="https://cdn-images-1.medium.com/max/1816/1*LZ9BcuQ0MxFcyUJuSWHopw.png" width="1816"></a></p><p class="medium-feed-snippet">Barely noticed, Claude Code replaced the simple to-do lists it used to generate with a more structured Tasks feature.</p><p class="medium-feed-link"><a href="https://medium.com/@dlaytonj2/claude-code-tasks-when-claude-code-plans-the-work-and-then-does-it-43f98f5113cd?source=rss-fce655d290ee------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@dlaytonj2/claude-code-tasks-when-claude-code-plans-the-work-and-then-does-it-43f98f5113cd?source=rss-fce655d290ee------2</link>
            <guid isPermaLink="false">https://medium.com/p/43f98f5113cd</guid>
            <category><![CDATA[ai-coding]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[openai]]></category>
            <category><![CDATA[data]]></category>
            <category><![CDATA[claude-code]]></category>
            <dc:creator><![CDATA[Dennis Layton]]></dc:creator>
            <pubDate>Sun, 25 Jan 2026 20:53:17 GMT</pubDate>
            <atom:updated>2026-01-25T20:53:17.856Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Long-Running AI Tasks: How to Keep Context Without Keeping the Entire Transcript]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@dlaytonj2/long-running-ai-tasks-how-to-keep-context-without-keeping-the-entire-transcript-10b9e11a0370?source=rss-fce655d290ee------2"><img src="https://cdn-images-1.medium.com/max/1232/1*s5V1msU-GqfHenRD0smr8A.png" width="1232"></a></p><p class="medium-feed-snippet">Introduction</p><p class="medium-feed-link"><a href="https://medium.com/@dlaytonj2/long-running-ai-tasks-how-to-keep-context-without-keeping-the-entire-transcript-10b9e11a0370?source=rss-fce655d290ee------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@dlaytonj2/long-running-ai-tasks-how-to-keep-context-without-keeping-the-entire-transcript-10b9e11a0370?source=rss-fce655d290ee------2</link>
            <guid isPermaLink="false">https://medium.com/p/10b9e11a0370</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[vibe-coding]]></category>
            <category><![CDATA[data]]></category>
            <category><![CDATA[openai]]></category>
            <category><![CDATA[chatgpt]]></category>
            <dc:creator><![CDATA[Dennis Layton]]></dc:creator>
            <pubDate>Sun, 18 Jan 2026 02:34:22 GMT</pubDate>
            <atom:updated>2026-01-18T02:34:22.175Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Conversation State Management: Three Approaches]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@dlaytonj2/conversation-state-management-three-approaches-63591aaebf95?source=rss-fce655d290ee------2"><img src="https://cdn-images-1.medium.com/max/1232/1*zYlTEe7OqNLPsQVwto-TUg.png" width="1232"></a></p><p class="medium-feed-snippet">OpenAI&#x2019;s Responses API is a low-level, stateless-by-default interface for generating model outputs: you send input, optionally reference&#x2026;</p><p class="medium-feed-link"><a href="https://medium.com/@dlaytonj2/conversation-state-management-three-approaches-63591aaebf95?source=rss-fce655d290ee------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@dlaytonj2/conversation-state-management-three-approaches-63591aaebf95?source=rss-fce655d290ee------2</link>
            <guid isPermaLink="false">https://medium.com/p/63591aaebf95</guid>
            <category><![CDATA[vibe-coding]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[openai]]></category>
            <category><![CDATA[chatgpt]]></category>
            <category><![CDATA[data]]></category>
            <dc:creator><![CDATA[Dennis Layton]]></dc:creator>
            <pubDate>Thu, 15 Jan 2026 02:09:11 GMT</pubDate>
            <atom:updated>2026-01-15T02:09:11.328Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Enterprise AI Adoption: Beyond the Benchmarks]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@dlaytonj2/enterprise-ai-adoption-beyond-the-benchmarks-47ed0660c584?source=rss-fce655d290ee------2"><img src="https://cdn-images-1.medium.com/max/1280/1*SOUZuT2VOvKyjFsf7FH6EQ.png" width="1280"></a></p><p class="medium-feed-snippet">At the end of 2025, one thing stands out to me: enterprise adoption of generative AI is still moving slower than many expected, and the&#x2026;</p><p class="medium-feed-link"><a href="https://medium.com/@dlaytonj2/enterprise-ai-adoption-beyond-the-benchmarks-47ed0660c584?source=rss-fce655d290ee------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@dlaytonj2/enterprise-ai-adoption-beyond-the-benchmarks-47ed0660c584?source=rss-fce655d290ee------2</link>
            <guid isPermaLink="false">https://medium.com/p/47ed0660c584</guid>
            <category><![CDATA[data]]></category>
            <category><![CDATA[chatgpt]]></category>
            <category><![CDATA[vibe-coding]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[openai]]></category>
            <dc:creator><![CDATA[Dennis Layton]]></dc:creator>
            <pubDate>Thu, 15 Jan 2026 02:02:54 GMT</pubDate>
            <atom:updated>2026-01-15T02:02:54.950Z</atom:updated>
        </item>
    </channel>
</rss>