<?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 Alexandra Leloup on Medium]]></title>
        <description><![CDATA[Stories by Alexandra Leloup on Medium]]></description>
        <link>https://medium.com/@alexandraleloup?source=rss-458a5025fdbf------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*4mN6RhxSRSc543nVZg2j7g.jpeg</url>
            <title>Stories by Alexandra Leloup on Medium</title>
            <link>https://medium.com/@alexandraleloup?source=rss-458a5025fdbf------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 21:08:52 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@alexandraleloup/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[Vibe Coding Won’t Kill PRDs — Or Discovery]]></title>
            <link>https://medium.com/@alexandraleloup/vibe-coding-wont-kill-prds-or-discovery-ca369d3f8fff?source=rss-458a5025fdbf------2</link>
            <guid isPermaLink="false">https://medium.com/p/ca369d3f8fff</guid>
            <category><![CDATA[vibe-coding]]></category>
            <category><![CDATA[product-development]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[product-design]]></category>
            <category><![CDATA[product-management]]></category>
            <dc:creator><![CDATA[Alexandra Leloup]]></dc:creator>
            <pubDate>Tue, 12 Aug 2025 08:29:34 GMT</pubDate>
            <atom:updated>2025-08-12T13:09:03.983Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gr2-KHEtQcR0e-qiIk3OZA.png" /></figure><h3>Vibe Coding Won’t Kill PRDs — Or Discovery</h3><p><em>AI can speed up building, but clarity and memory still win. The best product teams balance thinking, making, and documenting.</em></p><h3>TL;DR</h3><p>Vibe coding (AI-assisted building) is a rocket for <strong>solutioning and delivery</strong>. It does not replace <strong>discovery</strong> or the need to write down <strong>why, what, and how</strong>.</p><p>Your prompt is a micro-PRD; your PRD is the memory. Keep all three: <strong>thinking, making, and remembering</strong>.</p><h3>The Debate (and the False Choice)</h3><p>After the recent wave of posts about <em>vibe coding</em>, a lot of people jumped to:</p><blockquote><em>“Docs are dead. PRDs are dead. Just build.”</em></blockquote><p>Cool. <strong>Also wrong.</strong></p><p>Vibe coding is phenomenal for speed. It collapses the time from “idea” to “something you can click.” But speed without clarity is just <strong>fast confusion</strong>.</p><p>Discovery still decides <em>what matters</em>, <em>why now</em>, and <em>how to de-risk</em>.<br> Documentation still keeps the <strong>institutional memory</strong> so we don’t redo the same arguments every quarter or ship regressions with a fresh coat of hype.</p><h3>Prompts Are Micro-PRDs</h3><p>If you can’t explain the problem and constraints to yourself, you won’t explain them to an LLM.</p><p>Prompts are <strong>compressed PRDs</strong>:</p><ul><li><strong>Why</strong>: the user/job, the outcome you’re chasing.</li><li><strong>What</strong>: scope, constraints, edge cases.</li><li><strong>How</strong>: architecture hints, integration points, acceptance criteria.</li></ul><blockquote><em>Garbage in → garbage out… just faster and with nicer code comments.</em></blockquote><h3>A Story From the “Trio” Era</h3><p>Earlier in my career as a PM, the <strong>trio</strong> (PM, Designer, Developer) was the heartbeat of the work.</p><p>I would literally sit between my product designer and my lead developer. I’d share what I’d learned from customers — problems, opportunities, constraints — and the <strong>designer</strong> would sketch flows <em>as we talked (even sometimes argue but that tension was real creativity!)</em>. Meanwhile, the <strong>developer</strong> started coding immediately.</p><p>He was a machine — a real coding bulldozer, point him at a problem and you could hear the engine start (and the best part? He’d actually been a machine engineer before. You can’t make this stuff up. :-D)</p><p>As we kept talking (or argue), <strong>the solution came to life</strong> in real time. Within hours we had something we could <em>show</em> users.</p><p>And guess what? People react to working product faster than to slides. <strong>An image is worth a thousand words; a prototype is worth ten meetings.</strong></p><p>That loop was pure fire.</p><p>And still… I wrote the <strong>PRD</strong>.<br>Not because I loved paperwork, but because:</p><ul><li>It captured <strong>why</strong> we built it this way,</li><li>The <strong>trade-offs</strong> we made under pressure,</li><li>The <strong>acceptance criteria</strong> and open risks,</li><li>And because our <strong>QA team in India</strong> needed a clear source of truth to know <em>what to test</em> and <em>how</em>.</li></ul><p>People doing the building are not always in the room — and won’t always be there six months later.</p><p>The <strong>memory</strong> still matters.</p><h3>Vibe Coding Doesn’t Replace the Trio</h3><p>There’s a misconception that with AI-assisted development, we can do more with fewer roles — maybe even replace some.</p><p>That’s not how great products happen.</p><ul><li><strong>The PM</strong> still frames the <em>why</em> — the context, the problem, the risks, the bets. Without that, AI just builds faster in the wrong direction.</li><li><strong>The Designer</strong> still shapes the <em>how it feels</em> — flows, interactions, and the user experience details that turn “functioning” into “delightful.” AI can help produce mockups, but it doesn’t own taste or context.</li><li><strong>The Developer</strong> still makes it <em>real</em> — deciding how it scales, integrates, and performs. Even with AI writing parts of the code, human judgment calls remain at the core.</li></ul><blockquote><em>Vibe coding is like strapping a turbocharger to this trio.<br> It accelerates delivery, but it doesn’t replace the driver, the navigator, or the one designing the map.</em></blockquote><p>Without all three, you risk moving fast but delivering something nobody loves, uses, or even understands.</p><h3>What Vibe Coding Actually Changes</h3><ul><li><strong>Speeds up solution exploration</strong>: Your first prototype arrives in hours, not weeks.</li><li><strong>Improves stakeholder alignment</strong>: A demo is a better debate starter than a doc.</li><li><strong>Shortens the feedback loop</strong>: Users can touch it sooner, which crushes fantasy roadmaps.</li></ul><p>What vibe coding <strong>doesn’t</strong> change:</p><ul><li>The need for <strong>problem clarity</strong>.</li><li>The need to <strong>de-risk</strong> before you hard-commit.</li><li>The need to <strong>remember</strong> decisions, context, and constraints.</li></ul><h3>“But PRDs Are Slow and Heavy…”</h3><p>They can be. That’s a process smell, not an argument to forget <strong>why you built what you built</strong>.</p><p>If your PRDs read like a legal thriller written by a committee, fix the format — not the concept.</p><h3>A PRD for the AI Era (Lean and Useful)</h3><p>Keep it to what future-you and not-in-the-room teammates actually need:</p><ol><li><strong>Problem &amp; outcome</strong> (1–2 paragraphs): who hurts, how we’ll know it got better.</li><li><strong>Scope &amp; constraints</strong>: must-haves, nice-to-haves, things we’re explicitly <em>not</em> doing.</li><li><strong>Risks &amp; assumptions</strong>: the bets we’re making (call them out boldly).</li><li><strong>Acceptance criteria</strong>: black-and-white checks, incl. edge cases.</li><li><strong>Decision log</strong>: 5 bullets capturing key trade-offs (date + owner).</li><li><strong>Appendix: Prompts &amp; prototypes</strong>: link the prompts you used and the artifacts generated.</li><li><strong>QA notes</strong>: test strategy highlights, known gaps, and “watch-outs.”</li></ol><blockquote><em>Timebox it. One owner. No novel-length PDFs. </em><strong><em>Precision over prose.</em></strong></blockquote><h3>When You Skip Documentation, Here’s What Breaks</h3><ul><li><strong>Onboarding</strong>: New folks reverse-engineer your product by reading code and Slack archaeology.</li><li><strong>Post-mortems</strong>: You debate ghosts — no one remembers why a constraint exists.</li><li><strong>Scale &amp; compliance</strong>: Design, QA, and partners guess. Legal and security wing it.</li><li><strong>Roadmap sanity</strong>: Without history, every quarter re-litigates the same decisions.</li></ul><p>Speed is a feature. <br><strong>Repeatable speed</strong> is an advantage. <br>Repeatable speed needs memory.</p><h3>Discovery x Delivery x Documentation = Durable Velocity</h3><p>Here’s the real model:</p><ul><li><strong>Discovery is the thinking</strong> (clarity, bets, risk).</li><li><strong>Vibe coding is the making</strong> (speed, evidence, iteration).</li><li><strong>PRDs are the memory</strong> (alignment, continuity, quality).</li></ul><p>Kill any one of those and you’ll move fast until you don’t.</p><h3>The Punchline</h3><p>Vibe coding is not the end of PRDs. It’s the end of <strong>excuses</strong>.</p><p>If we can spin up working product in hours, we can spend the saved time being clearer thinkers and better storytellers of our own decisions.</p><p>Prompts are the new micro-PRDs. PRDs are still the memory. <br>And the trio — whether physically co-located or distributed — still wins when we keep talking, building, and documenting in the same breath.</p><blockquote><strong><em>Now, if you’ll excuse me, I’m going to see how we can inject that bulldozer energy across the org ;-) Speed is great — but speed with purpose is unbeatable.</em></strong></blockquote><p><em>This article reflects my personal views and is based on my professional experience. All examples are my own, and any resemblance to other works is coincidental.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ca369d3f8fff" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>