Noord is a fund and a studio. People assume that means two businesses sharing an office, and for the first year I would have struggled to argue otherwise. It is one thing, and the studio half is the part that does the underwriting.
The setup: we take paid brand and product work with early companies, and separately we write small checks (small enough that the work is the larger relationship, which matters later). The work comes first, it is priced like work, and it is useful whether or not anything else follows. Some of those engagements turn into investments. Most do not.
I did not design this on purpose. It came out of not being able to tell, from pitch meetings, which teams could actually build the thing they were describing.
What a pitch meeting cannot tell you
A pitch is a performance of clarity, and founders get better at it with repetition, which means the signal degrades exactly as the company gets more fundable.
You can assess the market from a deck. You can assess the logic. What you cannot assess is the thing that actually determines the outcome, which is how this specific group of people behaves when the plan meets something it did not anticipate. Every investor knows this and most compensate with references and reputation, which are useful and are also a description of the past.
I spent a year as an executive in residence at a seed fund in Amsterdam, running product and design reviews across a portfolio, and that was the first time I could see the gap plainly. I would sit in a partner meeting where a company looked strong on every measure you could see, then spend two hours with the product and find that nobody on the team had used it in a month. Neither observation was wrong. They were measuring different things, and only one of them predicted anything.
What three weeks of paid work tells you
Working with a team, for money, on something they need shipped, surfaces things that no amount of meetings will.
How a decision actually gets made. Not the org chart version. You send work back and watch what happens: whether one person decides, whether it goes quiet for four days, whether the founder overrules a specialist on the specialist's own ground and whether the specialist stays.
Who really holds the product. There is usually one person who can answer any question about how the thing works, and their position in the hierarchy tells you a great deal. When that person is a founder, the company is early and coherent. When it is a contractor, you have found the actual risk in the business.
What happens when we disagree with them. This is the one I weight most heavily. We will be wrong about something inside the first fortnight, because we do not know their customer yet. Teams that argue back with evidence survive being wrong later. A team that accepts a bad recommendation from a vendor they are paying is a team that will accept a worse one from an investor they need.
A team that accepts a bad recommendation from a vendor is a team that will accept a worse one from an investor.
The things I have learned to check
Some of this predates Noord and comes from my own screen, which I wrote down years ago and have edited less than I expected to. Proven value in an existing space, so somebody is already paying for something adjacent. Enough fragmentation that no monopoly has settled. A moat that is technical, meaning better experience and faster growth rather than one or the other. Incumbents who are mid to late stage and low tech. Nothing that is the Uber of something, no direct-to-consumer category creation, and no email, calendar or to-do application.
The studio adds a second list, and it is mostly about tells.
Fully outsourced engineering is the one I do not think I have ever seen work at this stage. Not because agencies are bad, but because the company has placed its only compounding asset outside its own walls, and every subsequent decision is priced accordingly.
A brand that has outrun the product is a softer signal and a common one. Beautiful identity, confident site, and then an application assembled from off-the-shelf components with the branding nowhere in it (usually not even in the sidebar). That gap tells you where the founder's attention goes and what they think the hard part is.
Then the measurements, which I have written about elsewhere and keep running into here: a team counting active users where one session counts as a user, or quoting a retention number with no window attached. What I am reading is not the number itself but whether anybody on the team has ever tried to break it.
The conflict, stated plainly
You cannot write about this arrangement honestly without naming the problem in it. We are a vendor to companies we may want to invest in, which gives us an incentive to be agreeable during exactly the period we are supposed to be forming a judgment. And the reverse: a founder who knows an investment might follow has an incentive to manage us rather than use us.
What we do about it is narrow, and I'm not certain it solves the thing. The work is priced and scoped as work, with no discount and no equity component, so the engagement stands on its own. We say early that an investment is not the expected outcome, because it is not. And I try to give the recommendation I would give if no investment existed, which is the only part of this that requires any discipline, because the temptation to soften a hard piece of feedback for a company you like is real and I have felt it.
The honest defense is that the alternative is worse. The standard arrangement, where you form a view from meetings and references and then wire money, has its own incentive problem: everyone involved is optimizing for the meeting.
What it costs
This does not scale, and I want to be clear that I know it.
We see far fewer companies than a fund running a normal process. Each engagement takes real weeks from people who could be deploying capital instead. You cannot do any of it from a spreadsheet, and it rules out entire categories where we have no ability to do the work. When we passed on a company recently, the honest reason was that we invest earlier than they were and had no depth in their space, and both halves of that sentence are constraints this model creates rather than removes.
What we get for it is a small number of decisions made with information that pitch meetings do not contain. For a fund of our size, that trade is obviously correct. For a larger one I doubt it is available at all.
I run Noord, an early-stage fund and studio in Amsterdam, and spent a year before that as an executive in residence at a seed fund here. No portfolio company or client is identifiable from anything above, deliberately. The counter-argument I take most seriously: this is a strategy that a small fund can run and a good one cannot, and the returns in venture are driven by access to a handful of companies who will never need to buy design work from anybody. If that is right, the model selects for teams who need us, which is not the same as teams who will win.
© 2026 Renato Valdés-Olmos