What we care about
Projectous exists to help people coordinate work with more clarity, trust and humanity.
Software that runs a team can be built two ways. One helps the people doing the work see it clearly. The other helps somebody else watch them. Those pull in opposite directions more often than anyone admits, and when they do, we go the first way.
Saying so is cheap, so here is the part that is not. Everything below is specific enough that you could catch us breaking it.
Things we will not build
Not features we have not got round to. These are off the table, whatever a feature comparison says and whoever asks for them.
No screenshot monitoring
Nobody works better because software is watching their screen.
No keyboard activity scoring
Movement is not effort, and measuring it teaches people to move.
No public shaming dashboards
A league table changes behaviour. Never in the direction you wanted.
No manufactured urgency
If something is not actually on fire, we are not going to tell you it is.
How the AI in it behaves
Projectous uses a language model in six places: translating a chat message, answering a question about your finances, matching columns when you import a file, the support assistant, working out whether a reply unblocks a task, and the weekly summary. It is a tool the product uses, not a colleague it is pretending to be.
It proposes, you decide
Nothing in Projectous acts on a model's answer on its own. Every one of them ends at a person choosing.
It does not grade anyone
No AI in here rates, ranks or scores the people in your account, and none of it builds a picture of how somebody works.
It gets the task, not the account
Each feature sends the content it needs to do its job. None of them is handed your account to read through.
Your work does not train it
The model is run by Anthropic on our behalf, and their terms say customer content is not used for training.
It says when it does not know
A confident answer to a question the data cannot settle is worse than no answer, so it is built to say so instead.
You can switch it off
Ask and we will turn the AI features off for your account. The rest of the product does not depend on them.
The binding version of the AI and data commitments is in the privacy policy. This page is where they are said in plain words; that page is where they are said in the ones that count.
Where this is still being worked out
Some of it is genuinely unsettled. How much guidance becomes nudging. When the software should interrupt somebody and when it should leave them alone. How to hold a team to what it agreed without flattening the reasons things slip. We would rather say that than pretend a list of principles answers it.
If something in the product reads like it went the wrong way, tell us. That is not a courtesy line — several of the rules above exist because somebody pushed back.