Twelve deliverables — the anatomy of a Moonlabs cohort portfolio

A reasonable question, asked by every parent and every sponsor we have spoken to: what does the £6,000 buy, in artefacts you can hold in your hand? The Academy's twelve-week curriculum lays out the schedule. This essay zooms in on the output — the twelve artefacts a graduate walks away with — and explains why each one is the one we chose to make load-bearing.

The principle behind the choice of artefacts is simple: each deliverable proves the underlying skill in a way a transcript cannot. The deliverable is the receipt. If a student can hold the artefact up and explain how they made it and what they learned doing so, the skill is real. The framework has been borrowed and tightened from the best cohort programmes we have observed, plus eighteen months of running operator-grade work alongside non-cohort hires.

1. A deployed, monitored product on a real domain

Not a Replit prototype. Not a Streamlit demo. A live URL on a real domain, with Sentry wired in from week one, sitting behind real auth and real billing rails. The artefact is the URL itself; the proof is you can give it to a stranger and they can use it.

We considered alternatives — a recorded demo video, a GitHub repo with a comprehensive README — and rejected them. Both can be faked. The deployed product, alive at a real domain you have paid for, cannot.

2. A buyer-focused landing page

Most landing pages teach themselves to be marketing pages over time — too many feature lists, too much social proof theatre, no clear ask. We require a sales document with five sections: what, for whom, why, evidence, ask. By week two the cohort's landing pages have stripped out everything not earning its keep.

The artefact is the live landing page. The proof of the skill is the conversion rate moving when the copy moves.

3. Three event-instrumented dashboards

PostHog or Plausible, instrumented properly, with a documented event taxonomy. Three dashboards: top-of-funnel (landing-page conversion), in-product activation (signed-up users who actually used the thing), and revenue (people who paid). The cohort has visibility on each of those by week three.

We considered allowing Google Analytics. We did not. GA tells you a lot of things you do not need to know and almost none of the things you do. The Academy is opinionated about the stack; we teach the stack we run in production.

4. Five recorded discovery calls

Real calls. Real buyers (or candidate buyers). Recorded with permission, summarised in a one-page note. The discipline is that these calls happen by week three, not after the product is "ready". The output is the recordings plus the notes.

The proof of skill is what changes after the calls. Founders who do this exercise tend to rewrite their wedge document materially. Founders who skip it tend to ship the product they had in their head, not the product the customer wants.

5. Wedge document v2

The first wedge document gets written in week one. The second gets written in week four, with the conversation data from the five discovery calls woven in. If the second document looks identical to the first, something has gone wrong; we coach against it actively.

The artefact is a one-page document. The proof of skill is how much it has moved from v1 to v2.

6. A bottoms-up market sizing

No top-down market sizing slides. We do not let the cohort write "this is a $42B market" on any slide. Instead: how many buyers exist, what each one pays, how many can be reached through which channel, what fraction realistically converts. The output is a one-page bottoms-up sizing model.

The proof is that an investor can sit across from the founder and stress-test each line of the sizing without finding a number that does not survive contact with reality.

7. A dilution model across three projected rounds

Most early-stage founders have never seen what their cap table looks like four rounds in. By week four, every Academy student has built and stress-tested a dilution model showing what happens at each stage. The artefact is a spreadsheet; the proof is that they can talk through it on a whiteboard without hesitation.

We considered teaching cap tables in week ten, just before demo day. We moved it to week four. The earlier the founder internalises the dilution arc, the better the round-shaping conversation goes at week ten.

8. A ten-slide deck (v3) and a populated data room

Not v1. v3. The deck has been rewritten in week five, rewritten again in week eight, and rewritten a third time in week eleven. The data room is populated with everything an investor would ask for at the end of a strong first call.

The proof of skill is that the deck survives the questions investors actually ask. We test this with our own network before demo day.

9. A financial model that has survived three operator stress-tests

Our financial-model template. Adapted to the student's company. Stress-tested by three independent operators (us plus one from our network). The model has unit economics, payback periods, and a sensitivity analysis on the two variables that matter most for the company.

The proof is that the founder can answer the standard set of investor questions about the model without having to check the spreadsheet. The model has lived in their head long enough.

10. A signed pilot, LOI, or paid trial

By week six. Real money or real intent on the table. We do not let the cohort move past week six without this — and we recalibrate the wedge if it has not landed by then.

The proof is the signed document. We have had cohorts where the signed pilot came in week three; we have had cohorts where it took until week eight. We have not had a graduating student arrive at demo day without one.

11. A SAFE template ready to issue

Standard YC-style SAFE, with the cap and discount populated by the student's own valuation work. Reviewed by a friendly lawyer before issuance. The artefact is a PDF; the proof is that the founder can explain every clause without consulting the lawyer.

We considered teaching priced equity rounds in the cohort. We moved that to "post-cohort if needed". At pre-seed, the SAFE is the right instrument for nineteen out of twenty companies, and learning the SAFE deeply is more useful than learning both shallowly.

12. A pitch you can run cold

Recorded multiple times across the cohort. Three versions: thirty seconds, two minutes, ten minutes. Body language coached. Anti-pitch drilled. By demo day, the founder can run any of the three versions in front of a stranger without warm-up.

The proof of skill is the demo-day pitch itself. The proof for the student is that they can run the pitch again the next month, and the month after, with strangers, without rehearsal. The pitch is now load-bearing, not theatre.

Why we chose these twelve

Two principles. First: each artefact is production-grade, not classroom-grade. If a senior operator picked any one of them up and reviewed it, the artefact should not embarrass the student.

Second: the twelve artefacts together compound into something larger than the sum. A landing page on its own is a marketing exercise. A landing page plus event-instrumented dashboards plus five recorded discovery calls plus a wedge document v2 is a running company. A pitch on its own is theatre. A pitch plus a deck plus a financial model plus a populated data room plus a SAFE template is an investable company.

The twelve weeks bake in the compounding. The deliverables make it visible.

What an applicant should hear in all this

If you are reading this and thinking about applying — the artefacts are the deal. Not the certificate, not the alumni network, not the badge on the LinkedIn profile. The artefacts. By demo day you will be holding twelve specific things that you built, with the operator-grade discipline of two senior operators in the room with you the whole way.

If those twelve things are what you want, the Summer 2026 cohort starts 15 June. Apply by Friday 5 June.

The Moonlabs Academy is a twelve-week, operator-led AI course in Derby. The curriculum deep-dive lists what is taught and shipped each week.