Everyone gives you their AI.owan is the one you own.
owan is an AI agent that lives on your computer. It runs on the AI you choose, and what it learns stays with you.
How is it different?
| Knows you | Runs on | Change AI | |
|---|---|---|---|
| Their agent | In their account | Their own model | Starts from zero |
| owan | On your Mac | ClaudeChatGPTYour own keyA local model | Keeps it all |
What is it built on?
- 01Guardrails
It asks before it sends, pays or deletes. Nothing that matters happens without your yes.
- 02Governance rules
A rule you set once holds on every task, whichever AI is doing the thinking.
- 03Encrypted vault
Your logins stay encrypted on your Mac. owan signs in for you, and the AI never sees a password.
- 04Sandbox
What it installs and runs stays in a sandbox, apart from the rest of your Mac.
- 05Model-agnostic
Claude, ChatGPT, your own key or a local model. Change it any time.
- 06Local-first memory
What it learns is kept on your Mac, and it is still there after every change of AI.
- 07Routing and caching
Each job goes to the model that fits it and reuses what owan already knows, so you spend less.
Questions.
Is owan a wrapper around someone else’s agent?
No. owan is its own engine, made in Europe. Its memory, its guardrails, its vault and the way it uses your Mac are ours. The AI model is the one part it borrows, and you choose which.
What leaves my Mac?
Only what a job needs, sent to the AI you chose, from your own account. What owan has learned, your logins and your rules stay on your Mac.
What happens when I change AI?
Nothing is lost. The memory lives with owan, not with the model, so a new AI starts with everything the last one knew.
Can it do something I did not ask for?
Not the things that matter. Before it sends, pays or deletes, it stops and asks you, and the rules you set hold on every task.
Does it work for a company?
Yes. For companies we set owan up with governance of their own, with policies for each team, an audit trail and admin controls. See owan for companies.