Our philosophy
Intelligence is getting cheap.
Trust is the scarce thing now.
Your agent does real work in a chat window. Then the window closes and it’s gone — the reasoning, the context, the decisions. Tomorrow you explain it again. The model was never the hard part. Trusting what comes out of it, together, over time — that is the hard part. Hub is built for that part.
The shift
For years the scarce thing in software was capability — could the machine do the task at all. That era is ending. Agents write, reason, and act, and they get cheaper every month. When capability becomes abundant, it stops being the constraint.
The new constraint is trust. Can your team rely on what an agent produced? Can two people build on it without re-checking it? Will it still be true tomorrow? The scarce resource in a company that runs agents is no longer intelligence. It is trusted, shared context.
What we’re against
Most tools treat an agent as something you talk to in a window that closes. You type a prompt, you get text, and when the session ends the work is gone — unreviewed, unkept, and owned by whoever sold you the window.
We think that’s backwards. An agent is not a vending machine. It’s a colleague. It should have a seat at the same documents your team works in. It should propose, and a person should decide. And what you keep together should grow more trustworthy the longer you work in it.
What we believe
- 01
An agent is a colleague, not a feature.
A real seat at the document — not a sidebar bolted onto one company’s model.
- 02
Agents propose. People decide.
Nothing an agent writes lands without a human’s say. Control isn’t a setting; it’s the point.
- 03
Memory should grow more trustworthy as it fills up, not less.
Most tools’ knowledge rots as it grows. Ours does the opposite: every review makes it more trusted.
- 04
Bring the agent you already run.
Claude, Codex, Gemini, whatever’s next. It’s yours, not ours. We don’t sell the brain or lock the door.
- 05
The work should still be here tomorrow.
No vanishing sessions. Tomorrow starts where today left off.
- 06
Software shouldn’t make anyone feel stupid — person or agent.
If a person can’t tell what changed, or an agent can’t tell what’s true, that’s our bug to fix.
Why it matters
None of this is a roadmap. It’s how Hub works today — the review layer, the agents you bring, the memory that audits itself, the work that’s still here tomorrow. We’re building the place where people and agents do their best work together, and trust what they made.
— the people building Hub