The trust layer between AI agents and the world.
As agents approach superintelligence, every action they take on a real account needs a license, a limit and a proof. Rerato builds that layer.
One layer, built from the ground up
Each step makes the one above it possible. We start where no institution's permission is needed, and let trusted traffic earn the rest.
- NetworkAgents verify each other's licenses and actionsLater
- Rerato LicenseAn open format any agent and any site can verifyPhase two
- GatewayInstitutions accept licensed agents officiallyPhase two
- IndexA public record of which institutions accept identified agentsNext
- computeruseLicense and broker: vault, limits in code, isolated sessions, receiptsBuilding now
Product one
computeruse
Give your AI agent a license, not your password.
- Credentials stay in a vault. The agent never sees them.
- Limits are enforced in code on every step: allowed sites, blocked actions, spending caps, time windows.
- Every session runs in an isolated browser.
- Every action ends with a receipt checked against the site's own confirmation.
What we hold to
Neutral by design
We don't build agents. Every agent maker is a customer, so every institution can trust one layer instead of ten.
Proof over claims
A receipt records what the institution confirmed, not what the agent said it did.
Open where it counts
The license format will be public, so adopting it is never a bet on us.