Why we run an incubator, not an agency

Every dev shop in the country is about to discover their billable hour is worth half what it was last year. The smart ones will stop selling hours and start selling outcomes. And the very smartest will stop selling outcomes and start owning them.

That is the entire thesis behind Moonlabs.

The agency model is structurally broken

For twenty years the standard playbook for a technical operator who did not want to grind through a single startup looked like this. You set up a small consultancy. You billed somewhere between £600 and £1,500 a day. You hired a few good people, eventually a few not-quite-as-good people, and you ran a margin on the spread. If you were any good you eventually had the choice between staying small and excellent or scaling and watching the quality slip.

The reason the model worked is that good software was scarce, expensive to produce, and slow to ship. Clients did not have a credible alternative. If they wanted a working booking system, a payments flow, a custom CRM, the only real lever was hours from people who knew what they were doing.

That world ended in 2023.

The honest version of what has happened to the agency model in the last twenty-four months is that the cost of production has fallen through the floor while the cost of judgment has held perfectly steady. The middle layer of an agency — the £550-a-day mid-weight engineer, the £400-a-day project manager, the £350-a-day designer churning out the same dashboard for the fourth time this year — is being silently revalued in real time. The market just hasn’t got around to telling them yet.

Look at what is genuinely scarce in 2025: someone who can take a fuzzy commercial idea, decide which thing actually matters, ship a working version of it, get a buyer to pay for it, and raise institutional money on the back of it. That person is rare. That person was already rare in 2015. The difference now is that the cluster of skills around producing the code — the bit that used to consume eighty percent of the agency’s headcount — has been compressed into a handful of operators with the right tools.

So we are in the bizarre position where the most leveraged engineer on the planet is suddenly cheaper to hire than ever, and the agency model is still set up to charge the buyer based on the opposite assumption. It is a business priced for the wrong decade.

You can ride that down, or you can step off

There are three honest moves available to a technical operator who has been running an agency in 2025.

The first is to ride the curve down. Keep selling hours. Lower your rate every year as the floor falls out. Compete with offshore shops and AI-augmented two-person teams. This is what most agencies are doing, mostly because they have headcount to feed and clients who are not yet ready to be told the new price. It is a real strategy. It is also a slow death.

The second is to pivot to outcomes. Stop charging for time, start charging for the thing the client actually wanted. £40,000 for a working ops portal that hits these five metrics. £15,000 for a checkout that converts at this rate. This is much better. The margin is yours to keep when you ship faster than the price assumed, and AI has made it possible to ship dramatically faster than the price assumed. We have run this model. It works. It is a real business.

The third — the one we picked — is to stop selling at all and start owning the upside. Build the company with the founder, take a small equity slice for the technical foundation, take a fair fee on the capital raised on the back of it, and let the compounding do the rest. This is the incubator model.

The incubator model is not new. Y Combinator, Entrepreneur First, every accelerator under the sun. What is new is that for the first time in our careers an incubator can plausibly be run by operators rather than capital allocators. Two people, in the room, on the actual build, taking a small piece of every company they touch. That was not a workable business model when the build took six engineers and a year. It is a workable business model when the build takes two operators and twelve weeks.

Why “small equity, fair fee” is the right shape

The standard incubator slice is somewhere between 6% and 12%. We take 1% for the tech-for-equity. That number is deliberately low and deliberately controversial. The argument we get from other incubators is that we are leaving money on the table.

We disagree, for two reasons.

The first is selection. At 6% you have to fill the cohort to make the maths work. You will end up backing companies you would not back at 0%. We would rather pick two a year we genuinely believe in than twelve we are lukewarm on. 1% lets us stay honest. We can say no to nine in ten applicants without flinching at the dent in revenue, because the revenue does not depend on filling seats.

The second is alignment. 1% is small enough that we have to actually help the company succeed to make it worth our time. A 10% slice can carry a moderately useful incubator for years on the back of one accidental hit. A 1% slice can only carry an incubator that picks well, ships well, and stays in the founder’s corner long enough for the equity to be worth something. It forces good behaviour by construction.

Then on top of that we take 10% of the round we help close. This is the part that does the financial heavy lifting. Pitch, deck, financials, data room, warm introductions to the high-net-worth individuals who actually write the cheques. It is paid on success, not on slides, which means we only get paid when the round closes, which means we only work on rounds we genuinely believe will close.

That structure — small equity, success fee on capital — turns out to be unusually well aligned with what AI-native company building actually looks like. The build is fast and cheap, so we cannot charge for it the way an agency would. The fundraise is the part that genuinely takes operator time and operator network. So that is where we charge. The pricing follows the work.

What the agency people miss

We meet a lot of agency owners on a similar journey. The ones who are struggling to make the leap to incubator usually have one of two blockers.

The first is psychological. They are used to billing hours and seeing the money land in the same quarter. Equity feels like Monopoly money compared to a £40k retainer that pays Friday. This is a real concern in the first year, less of one in year two, and a non-issue by year four. The math is on the side of equity. The patience is the hard part.

The second is operator credibility. To run an incubator you have to actually be able to operate. You have to have raised money yourself, sold a product yourself, made a P&L yourself. Founders can smell the difference between someone who has done the thing and someone who has watched a lot of people do it. Most agency owners can ship code; not all of them can run a company. The ones who can are in a uniquely strong position right now.

If you are reading this and you are running an agency, the playbook is genuinely available. Pick the two or three founders in your client base you would back with your own money. Offer to do the build for a small equity slice plus a success fee on the next round. Be the operator they cannot get anywhere else. Stop charging for hours that are about to be worth half as much again next year.

What we tell founders

The thing we tell every founder we talk to is that, in 2025, the build is the cheapest part of the company. The market knows this. Investors know this. Buyers know this. The premium has moved entirely to the team that can decide what to build, who to build it for, and how to get them to pay.

Moonlabs is structured around that reality. We do not sell the build. We do the build because doing the build is how we learn enough about the company to be useful on the rest. The rest — the wedge, the pitch, the round, the introductions — is the actual product.

If we have done our job, by the time the company has a Series A in the bank, the build was the smallest line item in the story.

That is the only honest place left to stand.

James Freestone is co-founder of Moonlabs. He previously raised £5m+ for his own companies and is the operator behind Homemove, home.co.uk and homedata.co.uk. The Moonlabs Incubator runs at 1% tech-for-equity plus 10% on funds raised.