Why I Am Building a Venture Studio

I have spent roughly thirty years in service businesses. Strategy consulting. Financial advisory. Mergers and acquisitions. Corporate transformation. Building new ventures for other people's companies, and occasionally walking into boardrooms to explain why the new venture they loved last quarter should be shut down this one.

I would not trade any of it. The years taught me how industries actually work, as distinct from how they describe themselves. They taught me how to read a market from fragments, how to structure ambiguity before it structures you, how to sit in a room with twelve executives who each have a different answer and leave with a thirteenth that none of them brought in. They taught me, above all, how to distinguish between a problem that is real and a problem that is merely articulate.

The limitation was never the work. The limitation was the structure.

At the end of every engagement, I handed over the keys and walked away. The value I helped create belonged to someone else. The judgment I accumulated compounded only inside my own head, not inside any asset on any balance sheet. Every new client was a new cold start. Revenue scaled linearly with hours. Time, I have come to understand, is the one input that refuses to compound.

The window AI has opened

We are in the early part of a platform transition that is being widely misunderstood. Some people see AI as a better search engine. Others see it as a faster summariser, or a smarter autocomplete. These framings are not exactly wrong. They are merely too small to be useful. What AI actually does, at its most valuable, is connect ideas across bodies of knowledge that no single person could hold in their head at once, and compress the distance between a question and a defensible answer.

The more accurate framing is this: AI has collapsed the cost of building software. A two-person team with the right tooling and the right judgment now ships, in a week, what used to require a team of eight and a quarter. Prototypes that demanded a sprint now compile in hours. Market research that consumed an analyst for a month now arrives in days, with footnotes.

Paul David showed that electricity took thirty years to deliver its promised productivity gains to American manufacturing. Early adopters bolted electric motors onto factory layouts designed for steam power and waited for the numbers to improve. The numbers did not improve until a new generation of operators redesigned the work from first principles. Robert Solow captured the same lag for computing in 1987. "You can see the computer age everywhere," he wrote, "but in the productivity statistics." That paradox persisted for nearly two decades.

AI is in the same place now. The tools have arrived. The operational redesign has not. That is the window.

And here is the thing about windows. They close.

Why a studio, not a single bet

The instinctive founder move, once you see the window, is to pick one idea and run at it. Raise a round, assemble a team, commit the next three years to a single thesis. It is the default playbook, and for certain people and certain ideas, it remains the right answer.

For me, it is the wrong one.

Thirty years of watching ventures fail has taught me something uncomfortable about first ideas. They are usually wrong. Not slightly off-axis. Structurally wrong. The market does not care how hard you push. The market cares whether the problem you are solving is the problem anyone has. Most of the time, the founder has mistaken an articulate problem for a real one, or has picked a real problem that nobody will pay to solve, or has picked a problem worth paying for and built the wrong thing to solve it.

The only way to find out is to test. Early. Cheaply. In the open. And to be willing to stop.

A single-idea founder can test, but they cannot easily stop. Sunk cost pulls them forward even when the evidence argues against it. I have seen this pattern enough times that I now consider it the default failure mode of serious, committed, intelligent people.

I include myself in that pattern. I spent five years building a hardware startup in the healthy drinks category. The logistics worked. The product worked. Thirty to forty percent of the employees in the offices where we piloted loved it. What we could not find was a buying model that converted pilot love into purchase commitment. The pressure to ship kept us from running the rival framings properly. I would do it differently today. That is part of why the studio exists.

A studio solves for this. A studio can test a thesis hard, read the signal, and move on. The infrastructure carries forward. The judgment compounds. The tooling compounds. The next venture starts with the prior venture's hard-won understanding baked in. This is the opposite of starting from scratch each time, which is what most founders do by accident and what most consultants do by structure.

What this studio is and is not

IdeaJetLab is not a fund. I do not take equity from clients. I charge for the work.

It is not an accelerator. There are no cohorts, no demo days, no twelve-week grind toward an artificial pitch event. We work with a small number of clients at a time, selectively, because selectivity is the product.

It is not an idea factory. Clients bring the vertical knowledge, the customer access, the conviction. We do not manufacture ideas on behalf of people who have none.

It is not a development shop. We do not sell hours of engineering capacity. Building, when it is the right next move, is a consequence of judgment rather than a substitute for it.

What we are building, with my partner Edgars Rozentals, is a company-building studio with a signature method called Wind Tunneling. The method takes a felt pain and a set of hypotheses, generates rival framings of the problem, runs each through the same convergent evidence test, and ends in a binary Kill-Or-Build Call. The output is either a plan robust enough to build against, or an explicit recommendation to stop before the client has burned a year of their life on the wrong premise.

We also build our own ventures. TerraNext. Twin Ladder. SoulMath. Each one began with a specific customer whose problem I already understood, and a structural reason AI changed the economics of solving it. Each one survived its own Wind Tunnel before it drew the first line of code.

The honest version

I am building a studio because the timing is right and the toolkit finally matches the ambition. Thirty years of services gave me the pattern recognition, the commercial instinct, and the cross-industry judgment that is difficult to acquire any other way. AI gave me the leverage to build at a speed that used to require a full engineering department and a year of runway. The platform is resetting, and operators who understand both the technology and the commercial reality are in a rare position to act.

The studio model is how I walk through the window. Not with one bet and a prayer, but with a system that treats failure as data, compounds every lesson, and collapses the time between hypothesis and evidence.

A last point, and then I will stop. Every founder, every operator, every serious project drifts eventually into its own narrow view of the world, shaped by the signals closest to hand. The signals closest to hand are usually not the ones that matter most. From time to time, every idea benefits from being put back in the tunnel. Not at the start. Throughout. That is what I am building this studio to do, for other people's ideas and for the ones I pick up next.

The organisations that will matter in the next decade are not the ones adopting AI fastest. They are the ones building the judgment to know which problems AI should solve, for whom, and why anyone would pay. That is the work. That is the studio.