An AI-first venture studio
We build with AI, deliberately.
We’re an AI-first lab, and we want people experimenting and building with AI to make our work meaningfully better. That ambition is the point. The care we take is narrow and specific: what’s in a prompt, and where it goes.
What we believe about building with AI
Six positions that bound how we adopt AI. Pro-AI without hype; careful in exactly one place — the data.
AI-first, on purpose.
We're a lab — building with AI is the work, not a side experiment. We want everyone reaching for it, testing what's possible, and moving fast. The early adoption is deliberate: we lean in because it makes the work better, not because it's new.
Know where the data goes.
What matters most is the data — what's inside a prompt, and where it travels once it leaves us. Before sending anything to a third-party model, it's worth a beat to ask what's in it and who can see it.
Using and wiring-in are different.
Reaching for an AI tool in your own work is one thing. Handing an autonomous agent standing access to shared systems — email, calendars, deals, money — is another. Both are welcome; the second just deserves a closer look before it's switched on.
Access is its own decision.
We trust each other — that part isn't in question. Trusting a colleague is simply separate from the access their tools are given. We grant standing access deliberately, one integration at a time, because the tricky parts often sit outside any single person's expertise.
Give agents their own identity.
When a tool starts acting on its own, we set it up the way we'd onboard a teammate — its own account, its own scoped permissions, its own trail — rather than running it on someone's personal access. An automated worker gets the access its job needs, and no more.
We set our own pace.
The lab moves quickly — adopting and automating early is part of the job. That pace is a choice we make here, and it doesn't automatically carry to every context we work in. Each one gets to decide how fast is right for it.
Using AI, vs. wiring AI in
Both columns start the same way: a person and an AI assistant. The difference is what happens next.
On the left, you use AI in your own work — drafting, research, code. Nothing new gains standing access to our systems, so it’s a clear go.
On the right, an autonomous agent is given a key to email, calendars, deals, or money. That key is what we look at — and we set the agent up with its own identity before it turns.
No new standing access. Use freely.
Standing access to systems. Review required.
Three more pillars
Stance is the why. These are the how — the boundary, the handbook, and the backlog the team shapes together.
This is the reference we own and grow — a curated handbook, not a wiki everyone edits.
Structure and editorial control are features, not bugs. It replaces the scattered docs with one place that stays current.