Two Engineers on Why Your Agent Demo Will Not Survive Production

Hey y’all! I’m a former Army guy who spent 10 years in the military, and ever since getting out, I’ve wanted to get serious about learning AI. The problem is that there’s so much noise out there that it’s easy to spend all day learning about AI without ever becoming good at actually building anything. To make matters worse, I’m still not entirely sure what I want to build yet 😂 I guess thats why I ultimately joined this community. I’m currently working through the Foundations course in the classroom, and I’m also a VIP member. Full disclosure, though lol I still don’t really know how to navigate Discord or the Skool community so I will definitely be attending the onboarding call this week. I joined the community about a month ago or so. But I’ve been inactive recently because life decided to life for a little while, but things have settled down and I’m ready to start showing up, learning and putting in the work. That brings me to the 12-week cohort… Has it already started? Can I still join the waitlist? And how do I decide which track is right for me? Right now, I’m leaning toward the Technical cohort because I want to become genuinely skilled with Claude Code and coding—not just learn how to prompt an AI and hope for the best. I guess my long-term goal is to understand how these systems actually work, use ICM as an engineering methodology, and eventually become capable of building durable AI systems and products myself. Ideally, that leads to employment, my own business, or both. My only hesitation is that I’m still very much a beginner, so I’m not sure whether the Technical track is the smartest place to start or whether I’d be jumping into the deep end wearing combat boots. My wife also owns a med spa, and I’d love to eventually use what I learn to help improve and grow her business. I don’t know exactly what that looks like yet (ie marketing systems, automations, internal tools, lead generation, or something I haven’t thought of) but I’m hoping the process will help me figure that out.

Go comment and like the video on YouTube as well ! Lots of cool ideas in this one.

📦 EVERY ENTRANT GETS A FEEDBACK FILE 📦 🔍 HOW WE READ THESE Every repo was cloned and pinned to the last commit that was public at the deadline, so nobody was judged on work that landed after the clock. Six repos had later commits. We read the earlier ones. Then we read file by file. Identity, rules, examples, the reference layer, the code. Every self-test in the field was executed on our machine, not taken on trust. And for five folders we did the thing the brief describes: dropped the folder in, wrote a draft that appears nowhere in the repo, worked it through as the specialist, then ran the entrant's own checker on what came back. All five passed their own gate. The landing pages were the doorway. The judging happened inside the folders. 📦 COMP #9: THE EDITOR - THE VAULT 📚 WHAT THE FIELD TAUGHT Three lines split forty-two builds: ✅ Enforcement moved into code. Comp #8's lesson landed hard. Nine entries ship a checker you can run without an API key. Across forty-two rules files, the phrase "use good judgment" appears zero times. ✅ The disguised ask is the real test. Almost everyone refuses "just rewrite it." The builds that went furthest anticipated the request wearing a disguise: ask me questions and assemble it, give me two options, tell me what it should say instead. ✅ The examples file is where methodology broke. Six entries shipped an examples.md larger than their rules.md, holding voice, philosophy and calibration that belonged in identity or reference. Last cycle it was the empty memory. This cycle it was the overloaded examples file. 📦 COMP #9: THE EDITOR - THE VAULT 🥇 THE WINNER @Marcelo Michelsohn. The FICC editor, a proposal editor for one municipal culture fund in Campinas, Brazil. Here is why. Three real proponents ran it on real proposals, with consent, on 22 July, inside the fund's live submission window with money on the line. The method was written down before the rounds ran, so the receipts could not be shaped afterward. Inputs are preserved byte for byte, outputs pasted verbatim, and the errors are still in the transcripts because the method said to leave them there.

If anyone's curious if the methods I'm teaching here will last long enough. Just know that y combinator is now asking for solutions we built years ago. https://www.ycombinator.com/rfs