This week @James Mackellar took the whole thing! Mayston is a research partner for UK neurophysios prepping stroke-rehab CPD. That's the kind of work where you walk in ready for the pushback or you don't walk in at all. So Mayston holds the debate for you. It won't open the discussion until it knows your current teaching position, it keeps the rival treatment schools apart so they don't blur into mush, and it hedges every claim on the GRADE evidence scale. Deepest folder in the comp. But here's why it won. This is productionize your opinion all over again. Mayston is built on a working UK physio's decades on the frontline. His dad. The follow-up calls are logged right there in the repo. James didn't research a persona, he built his father's life work into a tool, then shipped it like a launch. Landing page, a 60 second overview, a full walkthrough, annotated diagrams. Same lesson as last week, said a little different. YOUR expertise is the value. And when it's close to home, a son building his dad's craft into something other clinicians can actually use, you feel it. People trust that. You stay passionate because you care whether it's right. That's the whole game. - 📺 60-second overview: https://youtube.com/shorts/uDZyiZ-aD8w - 📺 Full walkthrough: https://youtu.be/gzHoSpZPWS0 - 💬 https://mayston.pages.dev - 🔗 https://github.com/JamesMack05/mayston 🎯 Honorable Mentions Six builds, no particular order, that made this so hard to call. 🇳🇱 @Arjen Stet , Lex Dutch worker-classification law, built for the exact enforcement wave the Belastingdienst just restarted. What got me, it refuses to cite a court ruling it hasn't verified, so no made up ECLI numbers, and it splits the sources that lead (statute, the High Council) from the ones that only signal (blogs, tax-authority posture). Every conclusion gets labeled settled, in motion, or not-yet-law. Bilingual EN/NL readme so the rest of us can follow it, plus a short intro video. This was my runner-up, and it was close.

A colleague and I were talking about learning design. I showed her Clief Notes. She mentioned her son and daughter both have dysgraphia. Writing is physically hard, slow, and exhausting for both of them. Their ideas are good. Getting them on a page is the problem. Then she said something that stuck. Her son had recently handed in a ChatGPT-written assessment without changing a single word. Not because he was lazy. Because the AI just did it for him, and he didn't know how to make it his own. She asked if I could build something different. Something that would help him do his own work without doing the work for him. So I built it. A Claude system prompt, not a folder structure yet, but a detailed one, with 10 task types: analytical writing, annotation, creative writing, short answer, practical reports, maths, recall, brainstorm, emails, and "not sure what this is." Each type gets a different level of scaffolding. Her daughter needed a different version entirely. Different year level, different profile, different adjustments. So now there are two. The first thing built in: voice. Writing is the barrier for them, not thinking. So when their typed answers start getting short, Claude notices and prompts them to switch. Talk instead of type. The ideas come out the same way, just through a different door. --- The rules that came from watching real sessions break down: Claude cannot start from nothing. The student has to give it something first, even three messy sentences. Claude can only extend, restructure, or shape what is already there. The seed has to be theirs. And if Claude adds anything interpretive that the student didn't generate, it has to say so out loud: "I added this part, keep it, change it, or cut it?" The work has to be defensible as theirs. That is the whole point. Don't ask "happy with that?" after every sentence. Just keep moving. One real check per paragraph at the end: "Read it through. Does it sound like you?" The first version asked this constantly. By the third paragraph she was typing "happy" without reading it.

🔥 @Bas Rosario took the top spot this week. 🇬🇧 Quick note first. Apologies for the late post today. Jake and I are deep in prep for London Tech Week coming up. Thanks for your patience while we get everything dialed in. 🙏 🎁 What Bas wins: He was already a Premium member, so we're converting his account. ✨ Free Premium for life. ✨ No more monthly charges. Ever. ⏰ The 7-day clock just reset. Next Monday we crown the next winner. Could be you. 🎯 How it works: - 📝 Post bad ass stuff - 💬 Help people in the comments - 🛠️ Share what you're building, what's working, what's breaking - ❤️ Engage with other members' posts The leaderboard tracks all of it. Whoever sits at #1 next Monday wins. 🎁 The prize, depending on where you're at: 🆓 Free member? You get lifetime Premium, free ⭐ Already Premium? We convert your Premium so you stop paying 👑 Already VIP? We convert your VIP so you stop paying Either way, you stop paying. Forever. Bas, congrats. You earned this. 👏

🎟️ PRIZE: FREE SEAT IN THE LYCEUM 🎟️ Pick your cohort. Technical, Business, or Creator. Your call. ---- 🇺🇸 Quick note first. This post is going up Today because we took Memorial Day off yesterday. To keep things fair, you've got until Sunday May 31st at 12:00 PM EST to submit. Same week of build time, just shifted. ---- 📋 THE CHALLENGE Build a folder-based AI researcher for a specific topic or industry. You pick the domain. This week's deliverable is one researcher folder that someone could drop into a Claude project and use as their personal research partner for whatever domain you've built it for. ---- 🎯 PICK YOUR DOMAIN The domain is yours. Pick something specific. Pick something you'd actually use. A few sparks to get you thinking: - 🏦 M&A activity in one industry (fintech, healthcare, defense) - ⚖️ Court cases in one area of law (employment, IP, immigration) - 🧬 Scientific research on one health condition or treatment - 🏘️ Real estate market dynamics in one city or asset class - 🥊 Competitive intelligence for one product category - 📜 Historical research on one period, place, or movement - 📚 Academic literature in one specific subfield - 📋 Regulatory developments in one sector - 📰 Journalism research on one beat (climate tech, AI policy, biotech funding) The more specific, the better. "Research assistant" is too broad. "M&A research analyst for early-stage fintech deals in the US and Europe" is right. ---- 🗂️ THE METHODOLOGY If this is your first comp, welcome. Here's what you need to know: This week (and every week) you're learning interpretable context methodology. Folders as architecture. Each file does one job well. Your researcher is a folder with five things: - 📄 identity.md (who the researcher is, what domain they cover) - 📐 rules.md (how they research) - 💬 examples.md (what good looks like) - 📚 reference/ (frameworks, source lists, key concepts) - 📖 README.md (how to use it)

Something that I get asked often when building projects for clients is ‘my session always maxes the tokens before I am done. Makes me restart each session.’ First thing I do is smile because I get to teach something which always makes me happy. Second thing I do is set up a Progress.md file and make it required to be read at every session start. This file has made my projects so much smoother and has actually saved me tokens often. If you would like to know more, comment below and we can have a chat about it ✌️👊