About
A small team, one conviction.
We're an independent team building the tool we want to use ourselves: an intelligence that lives on your computer and remembers your work. Nova is in private preview, and we build it in the open with our early users — this page is how we think and how we decide.
Why Nova exists
Software should remember.
The most capable tools you use forget you the moment you close them.
Every session starts over. Every context gets re-explained. We think that’s backwards — an intelligence meant to help with your work should understand your work, and keep understanding it over time.
So we’re building an intelligence layer that runs on your computer, remembers your projects, and works across your apps instead of inside one chat window. Not a smarter answer — continuity.
The full set of beliefs behind that is written down, plainly, on our principles page. This one is about how those beliefs turn into decisions.
How we decide
Four questions, before anything is built.
Every feature has to answer them. It’s how a small team stays pointed at the thing that matters instead of the thing that’s loudest.
Does it serve memory?
We build outward from continuity. If a feature doesn't help Nova remember, or use what it already knows, it waits behind the ones that do.
Can we prove it?
If we can't show it working on a real task you could reproduce yourself, it isn't ready to be claimed — however good the demo looks in our hands.
Is it honest to ship?
We'd rather ship less and label it truthfully than ship more and blur the line between what's real and what's intended.
Did someone actually need it?
Real use from early users outranks a clever idea on a whiteboard. The preview exists so that pull, not guesswork, sets the order.
How we evaluate a feature
What earns a place.
A feature ships when it clears all four — not three. The bar is deliberately plain, so we can’t talk ourselves past it.
- It can be demonstrated on a real task, unedited
- It respects the defaults — local-first, private, reversible
- Its status is marked honestly: in preview, in development, or vision
- It earns its complexity — it makes the core better, not just bigger
Proof over marketing
Show the work, don’t sell the idea.
The strongest claim we can make is a real task, unedited, that you can reproduce yourself. So we’d rather show one thing that genuinely works than describe ten that might. It’s slower, and it means the site has fewer bold promises than a launch deck — on purpose.
In practice it’s a discipline: nothing is called finished until it can be demonstrated; everything in between is marked in preview, in development, or vision. You should never have to guess which is which — and if a claim here ever feels softer than the software, tell us and we’ll fix it.
How feedback shapes it
Built with early users.
The preview is small on purpose. It’s a working relationship, not a launch list.
We onboard people in small groups and stay in close contact. When you write, the reply comes from a real person on the team — not a queue. Real use is what sets the order of what gets built next; a friction you hit in your actual work counts for more than a feature request in the abstract.
That feedback loop runs straight into the four questions above. Something an early user genuinely needs, that we can prove and ship honestly, moves to the front. You can follow where that leads on the roadmap.
Where it is
Built, and honest about it.
Nova is in private preview.
The foundation is built: memory, perception, file understanding, desktop control, and automation work today. What’s next is strengthening the reasoning and reliability on top of it — in deliberate steps.
You can follow the trajectory on the roadmap, and reach us by requesting access — every reply comes from a real person on the team.
See it do real work.
Nova is in private preview. Request access, and we’ll write when it’s your turn.
Request access