A day in the life of a Moonlabs cohort student

The single most useful question a prospective student has asked us in the run-up to the Summer 2026 cohort is "what does a day actually look like?". The website explains the curriculum. The FAQ covers the logistics. Neither of them quite gives you the texture. This post tries to.

This is Tuesday of week five. The cohort is partway through the build of their own products. The Coding pillar is in pgvector + retrieval; the Commercials pillar is in pricing; the Investment pillar is in early-stage cap table mechanics. The cohort is eleven students (one had pulled out two weeks earlier for personal reasons). The room is the upstairs space of our Derby office, eight minutes from the station.

08:45 — Arrival

The cohort starts arriving from quarter to nine. Coffee from the espresso machine in the kitchen. Some are catching up on the previous evening's homework; some are clearing inbox. Two students are debugging an issue from the previous afternoon's pair session on their laptops. The room is quiet but not silent.

We do not run a "stand-up". We tried it in the pilot cohort; the format added bureaucracy without surfacing what we needed. What replaced it is a daily intent: each student writes one sentence in the cohort Slack channel about what they will ship today. That happens by 08:55. James and Louis read them as they come in.

09:00 — Coding pillar block

Tuesday morning is a Coding pillar block. This week it is retrieval-augmented generation with pgvector. The format is forty-five minutes of teaching, fifteen minutes of Q&A, and ninety minutes of hands-on.

The teaching is operator-led, not academic. Louis runs through the actual schema we use in homedata.co.uk for vector storage, the actual queries we run, the actual gotchas we have hit. He shows real code on a real production database. The teaching is one example deep rather than one paradigm wide.

Hands-on follows immediately. Students pair up (we rotate pairs weekly) and apply what they have just seen to their own product's retrieval problem. There is no theoretical exercise; there is no toy dataset. Every student is implementing the technique against the actual data of the company they are building.

Louis and James circulate. Tap an open laptop, sit on the chair next to a student, ask what they are trying to solve, suggest a different approach, move on. The model is operator pair-programming with the cohort, not lecture-with-questions.

11:30 — Break

Fifteen minutes. Cohort tends to spill outside if the weather is OK; otherwise the kitchen. Conversations are usually about either the morning's technique or what someone's customer said in a sales call the previous afternoon. The mood by week five is significantly less "I am being taught" and more "we are colleagues working on parallel problems".

11:45 — Commercials pillar block

Tuesdays often pair Coding in the morning with Commercials in the late morning, because the cohort has the energy for it. This week is pricing.

James opens with the actual pricing exercise we ran when we launched Homemove. The original price point, the customer feedback, the price change, the resulting conversion-rate impact. Numbers on the screen, no hand-waving. Then he flips to each student's own pricing problem.

This block is roughly forty-five minutes of teaching plus forty-five minutes of each student rewrites their pricing page on the call. By the end, eight of the eleven students have a new price point on their landing page. Three of them have already messaged a prospective customer with the new pricing. One has had a yes back before lunch.

The Commercials pillar is the one most cohort applicants underestimate. It is where the highest-leverage compounding happens. We have written about it separately: see the academy three pillars for the rationale.

13:00 — Lunch

Catered, on the long table in the room next door. Tuesday is the day we eat together as a full cohort — including James and Louis. The other days are more flexible. We made this deliberate; the most productive hour for cross-pollination across the cohort is the one that does not look like work.

Topics over lunch: usually a mix of what each student's company is doing, occasional industry gossip, and roughly one debate per week about something the cohort disagrees on (the most heated to date was about whether Claude or Cursor was the more important tool for a 2026 builder; we agreed in the end that the question was poorly framed).

14:00 — Independent build block

The afternoon is the longest single block in the schedule: three hours of independent build. Students work on their own product. No pairs unless they actively ask for one. No teaching. The space is quiet, focused, occasional keyboard noise.

James and Louis are available but do not interrupt. The norm we have established: if a student wants help, they put a small physical card on their laptop screen ("pair please" on one side, "deep focus" on the other). The cards work. We use them ourselves.

Most days, two or three pair requests come up during this block. Sometimes it is a stuck-on-a-bug request; sometimes it is a "I am about to do something that feels load-bearing and I want a second pair of eyes" request. The latter is the higher-value pattern; we coach the cohort towards asking for it more often.

17:00 — Investment pillar block

The last hour of the day is reserved for the Investment pillar most Tuesdays and Thursdays. This week is cap table mechanics. James walks the cohort through the actual cap table from our Homemove seed round, the SAFEs we accepted, the conversions when the priced round closed. Real numbers, real founders' percentages, real dilution.

This is the pillar most cohort applicants are most nervous about and most leave fluent in. By the end of week ten, every student can read a term sheet without help and has built their own financial model that has been stress-tested in front of the room.

18:00 — Soft close

The day ends officially at 18:00. About half the cohort heads home; the other half stays in the office to continue building, often for another two to three hours. We do not enforce hours; we observe that the strongest students tend to stay later, but we do not interpret that as a moral signal. Some of the strongest students leave at 18:00 and protect their evening for sleep.

By week five, most evenings have become a mix of: a couple of students doing customer calls, a couple working on their landing page, a couple genuinely off-duty, and at least one in deep-focus build mode with headphones on.

What changes across the twelve weeks

Weeks one to three are heavily weighted towards Coding pillar teaching, because the cohort needs the production-grade defaults in place before they can build well. The Commercials and Investment pillars are light-touch in this window.

Weeks four to seven are the most balanced — the schedule above is representative. All three pillars are active. The cohort is shipping daily.

Weeks eight to ten skew towards Commercials and Investment. Coding is now mostly self-paced because the foundations are in place. Pitching, pricing, contracts, investor introductions, financial modelling — these dominate.

Weeks eleven and twelve are demo day prep. We work through every student's deck slide by slide. The Coding pillar is in maintenance mode for the cohort's own product, not in new teaching.

Why this format works

We did not invent any of this. The structure is borrowed from the best of the cohort programmes we have observed (YC, Antler, Entrepreneur First, Jointakeoff) and tightened against the failure modes we have seen in operator-distant teaching: too much theory, too little shipped product, too few real operators in the room day-to-day.

The compounding factor is James and Louis being in the room every day. There are programmes that are taught by faculty; there are programmes that bring in operators as guest speakers. The Moonlabs Academy is taught by the operators who run home.co.uk, Homemove, and homedata.co.uk full-time. The teaching is therefore one example deep rather than one framework wide. The bet is that one example deep produces operators who can ship; one framework wide produces students who can describe the framework.

If you are reading this and you are picturing the cohort as classroom-style, drop that mental model. It is closer to the operating floor of a small AI-native company that has eleven founders working on parallel companies, with the two senior operators of the company seated at the next desks.

How to apply

Summer 2026 applies before Friday 5 June 2026. The cohort starts Monday 15 June and runs to demo day on Thursday 3 September. Twelve seats. Tuition £6,000 (employer sponsorship available — see for employers).

The application form is at /apply. It takes about fifteen minutes. We answer every applicant personally within the week.

The Moonlabs Academy runs in Derby. The Academy page has the full curriculum and the for-parents page has the parent-facing explanation.