A Matt Harmsworth Company
Tell us about it out loud. Our engineers build a working version you can put in front of real people, in weeks. You judge it on a thing that runs, not a deck. Buy it and you own it outright.
We run a small number of builds at a time. Applications are read by a human.
How it works
Nothing gets built until a human from our team has sat with you and agreed it is worth building. That is the entire point of a first flight. You find out early, and you find out cheaply.
01
Apply, then talk. Not a form with thirty fields. A conversation, ten minutes, where you describe the problem the way you would describe it to someone in the pub.
You will be talking to an AI to start with, and we will tell you that up front because we are not in the business of pretending. It is patient, it is available at eleven at night, and it asks a better fourteenth question than a tired human does. Everything you say comes straight to us.
02
Within one working day, a written brief. What you said. What we think you actually need, which is occasionally not the same thing. What it would take to build. And the honest part: what could go wrong, and whether we think this is worth building at all.
Some people take that brief and go and build it themselves. Genuinely fine. It is still better than the spec document you were going to write.
03
A call with Matt and one of the engineers. No pitch deck. We go through the brief, you push back, we tell you where we think it is weak.
Sometimes the answer is no. Sometimes it is that this is an automation, not a software build, and here is how to do it yourself in an afternoon. We would rather say that than take the work.
Nothing is built before this call. Ever.
04
A fixed window. A working version, running on real data, in a real workflow, in front of you.
It will not be finished. First flights never are. Something will fall off. That is precisely what you are here to find out.
You have watched it work. You decide on evidence.
Buy it and it is yours. The code, the data, the intellectual property, assigned in writing. No price rises, no seat fees, no email one Tuesday telling you the platform is being retired.
Walk away and that is a real option, not a polite fiction. You will understand your own business better than you did a month ago.
The honest bit
Matt stands on stages telling business owners to go and build things themselves in Lovable and Replit. He means it. It is the cheapest way on earth to find out what you actually want, and you should absolutely do it.
Here is where it stops being enough.
The moment that thing touches customer data. The moment it takes a payment. The moment somebody's Tuesday morning depends on it working. That is the point where you need people who know what happens when it breaks, because it will break, and you will not know why.
We are not here to tell you that you did it wrong. Half our best conversations start with someone showing us something they built themselves that nearly works. You have already done the hardest part, which is proving to yourself that the idea is real.
We build the version you can run a business on. Same speed. Trained engineers on the final actions.
AI gives you the speed. People give you the judgement. Anyone selling you one of those is selling you half a thing.
Why this works
Matt built one of the first commercial drone businesses in the UK. When he started there were fewer than 100 licensed operators in the country. Four years later there were 8,000.
He watched the entire arc. The gold rush, the shiny kit, the money pouring into companies with no idea what problem they solved. He taught over 200 people to fly commercially. Roughly half of the ones who went off to start a business failed.
Every one of them was early. Early was not enough.
"New kit, the latest software, super solutions to problems you don't know exists yet. I call this, 'have drone find use case'."
Swap one word and that is the AI market in 2026. Have AI, find use case. Everybody buying the tool first and hunting for a reason afterwards.
The operators who survived the drone years did the opposite. They started with a problem somebody would pay to solve, got very good at that one thing, and owned it outright. Their application, their data, their client relationships. The ones who did not make it were reselling somebody else's kit and calling it a business.
That is why we build you something you own, and why we will tell you when your idea is not worth building.
<100
UK commercial drone operators when Matt started
8,000
Four years later
~50%
Of his students who started a business and failed
Proof
Two things we have built and run in a real business. Both started as a problem somebody was solving badly by hand.
We built these for our own company. They run it every day.
ROAVR Core
An internal operations board for a survey business. Boards, cards, folders and activity, replacing a pile of spreadsheets and group chats.
ROAVR Fly
A drone operations compliance system. Aircraft register, batteries, pilots, flight logs and an audit pack the regulator can read.
Ownership
Most AI products are a rented wrapper on somebody else's platform. The day the pricing changes or the API is retired, so is your business process.
When you buy the output of a first flight, you get all of it. Source code. Database. Documentation. Full assignment of intellectual property, in the contract, in plain English.
If you never speak to us again, it keeps working.
If you are in the second column, no hard feelings. Far better we find out now than six weeks in.
Not sure which column you are in? Book a clarity call and we will work it out together.
The backstory
Matt wrote a book about being early to a technology everyone said would change everything. It covers the good bits and the expensive bits. The kit that went obsolete, the crash on a Grand Tour shoot, a night spent asleep in the back of a pickup on Skye, and the several thousand businesses that arrived after him and mostly did not make it.
If you want to understand why we build the way we build, it is all in there.
Ten minutes. Tell us the thing you have been describing in meetings for two years. We will tell you honestly whether it is worth building.
Applications are read by a human. We run a small number of builds at a time.