Every studio website lists what it can do. Those lists are interchangeable: strategy, engineering, design, delivery. Nobody reads them and learns anything, because a capability claimed costs nothing to write.

Our studio page carries a different list — the engagements we turn down, stated in public, next to the ones we want. This article is the reasoning behind that list, line by line, and behind the general habit it belongs to: when a fact about how we work is checkable, we would rather publish it than perform around it.

A refusal is a claim you can be held to

A capability list makes promises about the future, which is exactly where a reader cannot check them. A refusal list makes a claim about the present: bring us this kind of work and we will say no, and you can test that today by bringing it.

That asymmetry is the whole point. A studio that says "we do product engineering" has told you nothing that distinguishes it. A studio that says "we will not take staff augmentation without product ownership" has told you something falsifiable — and has accepted a cost for saying it, because some of the work it just ruled out would have paid well.

Costly statements are the only ones worth reading. The rest is formatting.

Line one: staff augmentation without product ownership

The first refusal on the studio page is staff augmentation without product ownership. Renting out engineers is the easiest revenue a small studio can book: no scoping risk, no outcome to defend, an invoice per seat.

We rule it out because of what it does to judgment. A seat in someone else's process executes decisions made elsewhere; it is not responsible for whether the product works, only for whether the tickets closed. Everything our operating thesis claims — that the people shaping the work also build it, that nothing is translated through a layer that never carries the result — becomes untestable in that arrangement. The engagement would pay, and it would teach us nothing we could stand behind.

Ownership is not a preference here. It is the mechanism the rest of the model depends on.

Line two: delivery plans built around an immovable feature list

The second refusal is delivery plans built around an immovable feature list. This one reads as arrogance until you look at what a frozen list actually asserts: that everything worth knowing about the problem was known before the work began.

Building products does not go that way. The working loop on the studio page — frame, build, ship, compound — exists because the most valuable information arrives after shipping, and a plan that cannot absorb it converts learning into a contract violation. Under a frozen list, being right too late is indistinguishable from being wrong.

We ask for room to challenge scope before committing to output — it is listed as a fit criterion, not a courtesy. An engagement without that room has decided in advance that the builder's judgment is not part of what it is buying, and there are cheaper places to buy work without judgment.

Line three: fabricated proof and dark patterns

The third refusal is work that depends on fabricated proof or dark patterns. Declining it is not the interesting part — most people decline dishonest work when asked directly. The interesting part is what enforcing it on ourselves turned out to cost.

This website has no invented metrics, no testimonials, no named score. Those absences were design decisions, made when the honest version was noticeably worse for us than the fabricated one would have been. Two current examples, both checkable right now:

  • The ventures index says new ventures are being prepared, because none of our products is listed publicly yet. A placeholder portfolio was sitting in the codebase; we deleted it rather than dress the page.
  • The contact page publishes no email address, because the mailbox is not live yet. An address that cannot receive mail would look more professional and be a small lie; the page says the channel is being prepared instead.

A refusal you have never paid for is a slogan. These are the receipts.

Why publish any of this

The three lines could have stayed an internal policy and filtered the same work. Publishing them does two extra jobs.

It moves the mismatch earlier. A prospect holding a frozen feature list reads the studio page and does not write to us, which costs one potential invoice and saves both sides the slow, expensive version of the same discovery three weeks into a scoping call.

And it sets a price on our own consistency. A public refusal list is a standing invitation to catch us violating it. That pressure is useful — it is much harder to quietly take the convenient engagement when the page saying you won't is the second link in your own navigation.

The foundry model — operating our own products and taking client work on the same standard — only means something if the standard is visible enough to be checked. The refusal list is the visible part. The rest of the site, including its empty states, is the evidence that we apply it to ourselves first.