Coding is not the bottleneck anymore. Commercials are.

We have spent the better part of a year teaching student cohorts in Derby, and the pattern is now unmissable. Every twenty-year-old in the room can ship a working SaaS in a weekend. Almost none of them can sell it, price it, or raise on it.

That gap is the single most important thing happening in early-career technical education right now, and almost no one is teaching to it.

The shape of the bottleneck has moved

For most of the last forty years, the binding constraint on a young founder was technical. Could you actually build the thing. If you could, the rest of the journey — sales, pricing, fundraising — was downstream. You hired or co-founded those skills in. The build was the moat, because the build was hard.

The build is no longer hard. We watch this every week. A first-year computer science student, given a Cursor licence and a clear product brief, can ship something that would have taken our 2018 team a quarter. The plumbing is solved. The patterns are baked into the models. The boilerplate is free. The wedge that used to cost £200,000 of engineer time now costs a weekend.

This is wonderful, and it is also a problem, because the people coming out of universities and bootcamps have been trained for a world where the build was the bottleneck. They have spent three years getting good at the thing that no longer matters as much as it used to, and almost no time on the things that now matter more.

We see this in two specific failures.

Failure one: they cannot price. Ask a graduate engineer what to charge for the SaaS they just built and you will get one of three answers, all wrong. Whatever everyone else charges. I don’t know, £9 a month? Free, to get usage. The actual answer — work backwards from what the buyer would pay to make the pain go away — requires a kind of commercial empathy that nobody has taught them, because for forty years it was assumed they would learn it on the job from someone older.

Failure two: they cannot raise. Ask the same graduate to write the deck and you get an engineering document with diagrams of the architecture, screenshots of the product, and three slides at the back about “monetisation.” The actual deck — the one an investor will fund — looks almost nothing like that. It is a commercial argument that happens to mention the technology. Nobody has shown them what one looks like up close.

These failures are not because the students are bad. They are because the curriculum is from the wrong decade.

The three pillars

When we sat down to design the Moonlabs Academy curriculum, we tried to be honest about what an operator actually does in 2026. We landed on three pillars, and the course is structured so that every week of the twelve-week cohort touches all three.

Coding. Not “learn Python in twelve weeks.” Coding the way operators code in 2026: Cursor, Claude Code, agentic workflows, evals, deploy pipelines, observability. The actual stack we use every day to run real companies. The point is not to make the student a senior engineer. The point is to make them a competent commander of AI tools, which is a different skill and a more valuable one. We want a graduate who can sit in front of an empty repo on a Monday morning and have a working, deployed, observable product by Friday. That is the bar.

Commercials. This is the pillar most other courses skip entirely, and the one we think is the highest-leverage. Positioning. Pricing. The first ten sales calls. Contracts. Distribution. Margin. The brutal arithmetic of customer acquisition. We do not teach these as theory. We teach them by having the student price and sell their own project, with real prospects, under real conditions, with one of us in the room when it lands. A student who has fumbled their way through three discovery calls in week six is in a different universe of capability than one who has read a chapter on “sales fundamentals.”

Investment. The third pillar is how money actually moves in a tech company. How a round is structured. What investors look at. The financial model that survives diligence. The data room. The HNWI introduction. The follow-up email that converts. We teach this because every operator at the level we want them to reach will, at some point, have to either raise money or work next to someone who is raising money. Being illiterate in this skill is not survivable.

These three pillars are deliberately not equal in difficulty. Coding is the most teachable because the tools are the best. Commercials is the hardest because it requires reps that take time. Investment is the most pattern-rich because every round looks remarkably like every other round once you have seen ten of them. We weight the contact hours accordingly.

What we do not teach

It is worth being explicit about what is not in the curriculum, because the absences are also signal.

We do not teach computer science fundamentals beyond what is necessary to operate the tooling competently. There are excellent courses for that and they are not on our critical path.

We do not teach “design thinking.” We teach pricing and positioning, which are the things design thinking is supposed to be a proxy for.

We do not teach “entrepreneurship” as a generic concept. We teach you to ship one specific company that is actually possible to ship. The generalisable lessons fall out of the specific work.

We do not teach pitch coaching as a standalone skill. A pitch that holds together is downstream of a deal that holds together. We work on the deal.

The closest mental model is medicine. Doctors do not learn medicine by reading about it. They learn medicine by being on rotation with senior doctors, on real cases, getting feedback in real time. The Moonlabs Academy is on rotation. James and Louis are the senior doctors. The case is whatever each student is shipping that week.

Why this is going to be the new shape of technical education

The wedge for technical education over the next decade is not “teach more people to code.” That problem is being solved at an astonishing rate by the model companies themselves. Cursor and Claude Code are better teachers than most lecturers, in the narrow sense of teaching syntax and patterns. If your education business is built on being a better syntax teacher than Claude, it has a sell-by date.

The wedge is operator literacy. Producing graduates who can take a fuzzy idea and turn it into a real, sold, funded company. That is the work that AI does not yet automate, and the work that pays. A graduate who can do that commands a premium that has historically been reserved for people with ten years of post-MBA experience. The students we have run through the Derby cohorts have already been offered roles that, in the 2015 economy, would have been senior product hires.

We expect, frankly, that within five years most serious technical courses will be structured the way we have structured ours. We are not under any illusion that the three-pillar framing is uniquely insightful. It is just the obvious framing once you stop pretending that the build is still the bottleneck.

What this means if you are eighteen

If you are eighteen, or you are the parent of an eighteen-year-old, the takeaway is straightforward.

The single highest-leverage thing you can do with the next twelve weeks is not to grind through another Udemy course on React. It is to ship one thing, sell one thing, and learn what raising money actually feels like. The coding will follow. The coding is the easy bit now.

We built the Academy because the existing system is not set up to deliver that experience, and we have the operator infrastructure to deliver it ourselves. If that is the shape of education you are looking for, the application form is open, and we read every one personally.

The Moonlabs Academy is a twelve-week, cohort-based course taught by James Freestone and Louis O’Connell-Bristow. Twelve students per cohort. £6,000 tuition (with multiple funding routes available). Application is on motivation and fit, not on who pays.