Age of Builders. Coding stopped being a profession. It is a literacy now.
Two cohorts run in parallel for five evenings. The Builder cohort is for the operator, the founder, the writer, the consultant — anyone with a clear idea and an internet connection. By the last evening, you have shipped a real, working tool. The Engineer cohort is for founders, CTOs, and technical operators in companies investing seriously in AI: software is going to zero, and these five evenings are about building the moat that is not.
Same week. Same ethos. Different starting point.
Until late 2025, this workshop could not have existed. The agents were not good enough. As of now they are: a clear brief, a few honest iterations, and a thing that used to take a small engineering team takes an evening. The bottleneck has moved. It is no longer syntax. It is taste, specification, and the discipline to ship. Those three are what we teach, ruthlessly, in five evenings.
We run two cohorts in parallel because the rooms are fundamentally different. Pick the one that fits. We sometimes recommend the other one in our reply.
Builder cohort
For the operator running the spreadsheet that runs the business. For the founder with three ideas and no engineer yet. For the designer, writer, or researcher who has always been the bottleneck on their own work. For the consultant whose work would land harder as a small tool than a slide deck. For the person who has been told that software is for other people, and is finished with that.
You will not learn to code. You will learn what software is — and learn to brief an AI agent the way a senior PM briefs an engineer, so the agent does the writing without losing the plot. By the last evening you have shipped something you actually use, with at least one other person using it too. The kung-fu moment is somewhere around evening two.
Engineer cohort
For the founder building an AI-native company. For the CTO whose SaaS product is suddenly cloneable in a weekend. For the staff engineer at an enterprise that is investing seriously in AI and wants the bet to land. For the solo builder who needs the moat from day one because the next vibe-coder is six hours behind them.
The premise is harsh: software is going to zero. A competent product is now a weekend of prompt-craft, and the valuation premium of “we built it ourselves” is collapsing — pure-software companies are seeing 70 to 90 percent of their valuation compressed against this reality. The companies that survive the next decade will be the ones with moats AI cannot easily replicate — distribution, brand, community, data flywheels, trust, regulatory depth. The five evenings are about diagnosing where your real moat lives, building the AI-native company around it, and shipping the part of your work that compounds rather than commoditises.
From “I cannot build software” to “I just shipped” in five sittings.
Evening 01 — What software actually is
The hour that breaks the spell. Strip the magic. Every app you have used — Instagram, your bank, the tool you book flights with — is the same three things in different combinations: state (the data), logic (the rules), an interface(what the human sees). We build a tiny one from a single sentence: “I want a small tool that remembers every article I read and the one thing I took from each.” The three parts appear in front of you. By the end of the evening, you cannot un-see them in any app you open. This is the kung-fu moment. Everything afterwards is leverage.
Evening 02 — Thinking in data
What data is. What it is not. The four moves every app makes on it: find, sort, change, count. Spreadsheets are databases. A list of users is just rows. JSON is just a folder with labels. The body of knowledge that took a CS programme a year to teach, in a single evening, with the agent doing the typing and you doing the deciding. By the end of this evening, you can describe — out loud, in plain English — exactly the data your idea needs and the four things it should do with it.
Evening 03 — Briefing the agent
The new core skill is not syntax. It is specification. We teach you how to brief an agent the way a senior PM briefs an engineer, or a director briefs a designer. What it gets wrong, and why. How to give it just enough context. How to push back when it gives you the confident-sounding wrong thing. The taste move that separates a builder from a person typing into a chatbot. We work on three real briefs from the room and rewrite each one until the agent ships the right thing on the second try.
Evening 04 — The build loop
You write the brief. The agent writes the code. Something breaks. You read the error — actually read it, with the agent translating. You change one thing. You ship a tiny version. We run this loop live, three times, on three of your real projects, with the team in the room to unstick. By the end of the evening, the room has working things in it.
Evening 05 — Shipping
The smallest path from “works on my laptop” to “three people I know are using it.” Domains, deployment, sharing the link. The honest test of a built thing: is anyone using it? Each builder closes the workshop by demoing what they made and the first user it goes to that night. We do not stop until that user has opened it.
Software is going to zero. Build the moat that is not.
Session 01 — Software is going to zero
The strategic frame, in its harshest form. A competent SaaS product is now a weekend of prompt-craft. The valuation premium of “we built it ourselves” is collapsing 70 to 90 percent against this reality. The companies that thought their software was the moat are about to learn it never was. Each participant arrives with a one-paragraph description of their company or product. By the end of the evening, they have named — out loud, to the room — the part that is going to zero, and the part that is not.
Session 02 — Where the moats actually live now
The seven moats that survive in an AI-native world: distribution, brand, community and network effects, data flywheels, trust, regulatory depth, embedded relationships. We diagnose each participant's business against this list, hard. Most discover that the moat they thought they had is not the moat they actually have. By the end of the evening, every participant leaves with a brutal version of their own moat audit and a ranked list of what to invest in for the next year.
Session 03 — AI-native architecture
How to build the company so AI compounds the moat instead of eroding it. The inversion: stop building software as a product to charge for; start building it as the connective tissue between your moat and your customer. We walk real AI-native blueprints — the agent layer, the data layer, the human layer — and each participant maps their stack to one. The technical patterns under the hood (harness discipline, evals as code, MCP, retrieval that survives reality) get covered here, but as instruments of moat, not ends in themselves.
Session 04 — The data flywheel
The one software-shaped moat that survives AI commoditisation: proprietary data that gets richer the more your users engage. We work through the anatomy of a real data flywheel — the loop, the hardness, the unfair compounding — and the three failure modes that make most “data moat” claims hollow. Each participant designs or audits the flywheel that should anchor the next year of work.
Session 05 — Distribution, brand, trust at AI speed
The human moats AI cannot clone, accelerated by AI. How a brand becomes the default that AI agents recommend to other AI agents. How distribution compounds without the CAC death-spiral. Why trust is the highest-leverage asset in a world where every product is one prompt away from a clone. We close with each participant naming the 90 days of work that builds their non-software moat — and the part of their roadmap to delete, because AI now does it for free.
What both rooms share.
- Ruthless ignoring of minutiae. In five evenings we ignore everything that does not move you toward shipping — language wars, framework choices, computer-science fundamentals you can pick up in three weeks if you ever need them. The course is what you cannot pick up alone.
- The room is the curriculum. One-on-one, sometimes a pair or trio. We do not run large workshops. We run a conversation, with your real problem at the centre of it. What you take away is what cannot be picked up alone.
- You leave with something real by the last evening. Builder cohort: a deployed thing with at least one user. Engineer cohort: a strategic plan you can act on Monday — your moat audit, your AI-native architecture, the 90 days of work that compounds. Not a deck. Not a thesis paper.
- Taste over tools. The tools change every quarter. The taste — what good code looks like, what a good app feels like, what to push back on, what to accept — does not. We spend more time on this than the tutorials do, on purpose.
The right room.
Builder cohort
- Operators who maintain the actual spreadsheet that runs the business, and feel the seams in it
- Founders with three ideas and no engineer yet
- Designers, writers, and researchers who have always been the bottleneck on their own work
- Consultants who want their thinking to land as a working tool, not a deck
- Anyone who has been told they “are not a numbers person” and would like to call that bluff
You do not need a programming background. You do need a clear idea, a willingness to be wrong about it twice, and a laptop with one of the agents we use in the room running on it.
Engineer cohort
- Founders building AI-native companies — solo, small team, or seed-stage
- CTOs and VPEng at SaaS companies whose product is suddenly cloneable in a weekend
- Staff engineers and tech leads at enterprises betting seriously on AI and wanting the bet to land
- Operators making the architectural and strategic AI calls this year
- Solo builders who need the moat from day one, before the next vibe-coder catches up
You should be comfortable making decisions about where money and time get spent. The technical sessions assume some familiarity with shipping software, but the work of the cohort is strategic, not operational. We work in whatever language your project calls for — the agents handle the implementation; the decisions are yours.
The when, where, and how much.
| Format | Five evenings, 7–10 PM IST. Online. In-person option for enterprise cohorts. |
|---|---|
| Cohorts | Builder and Engineer run in parallel. Apply to whichever fits; we will sometimes recommend the other. |
| Next cohort | By application. We open dates as cohorts fill. |
| Cohort size | 1-on-1, or a pair or trio at most. We do not run group workshops. |
| Tools | Bring your own subscription to Claude Code, Codex, or Cursor — those are the agents we work with in the room. Bring your laptop. We work in whatever language your project calls for. |
| Engagement | Pricing on request. Enterprise engagements available. |
| Apply | One paragraph. Builder cohort: what you want to build, and why nobody has built it for you. Engineer cohort: the company or product you are betting on, and what you suspect is going to zero about it. |
Apply. We read every application.
We turn down roughly half. The room is small on purpose; we run this small.