Raidiam Connect
Put AI agents into production. Without losing control.
Your platforms now meet agents they hold no record of. Each platform resolves the agent through OpenID Federation, to an anchor it pins in Raidiam Connect, and decides for itself.
An agent
New to your platform. It signs its request.
1Asks
Your platform
The authorisation server in front of your APIs.
Client record None for this agent
2Resolves
3Signed answer
Raidiam Connect
The signed chain your platform resolves, ending at an anchor it pins.
- Who is this? Its own federation identifier, and the keys that speak for it.
- Who stands behind it? The operator that runs it, named on the record.
- What can it do? The types it may request, once each hop's policy is applied.
4Decides
Should I trust it here?
Your policy decides
The token carries the types requested, inside the authority's ceiling.
Trusted ecosystems Raidiam runs
- Open BankingBrasil
- Open Finance
- Central Bankof the U.A.E.
- ConnectID
- GetVerifiedNew Zealand
What changes when trust is shared
-
Know
Every agent you list, with its operator and its authority, on one record.
-
Connect
A new platform resolves the same signed statements as every other platform.
-
Control
Each platform applies its own policy to every agent it admits.
-
Change
Change it at the source. Each platform reads the change when its held answer expires.
Why believe it
Proven where trust is regulated.
Published by Raidiam for Raidiam Connect in production. Five of eight ecosystems are named above, each with hundreds of entities. Open Banking Brasil holds thousands of registered client applications carrying non human identities. This showcase is a separate estate, counted on the estate page. Open the showcase estate page
Built on open standards
- OpenID Federation 1.0 (open standard)
- OAuth 2.0 (open standard)
- X.509 and mTLS (open standard)
- SPIFFE (open standard)
The trust model is a published specification, not a Raidiam protocol.
Use cases
Start from the problem you have.
Keep your catalogue. Govern the agent.
Approving software and issuing it an identity are two decisions with two owners.
Resolved, not registered.
A platform meeting the agent for the first time admits it by resolving the authority.
Same software. Many environments.
One release. A governed identity in each environment.
Wherever it runs, proof sets the ceiling.
Where an agent runs decides what it can prove. What it can prove decides what it keeps.
Across the boundary. With trust.
They bring claims and evidence. You decide the effective authority.
When something changes. Access adapts.
Change authority once, and every consumer enforces it at its next read.
Your customer's assistant
Your own assistant. Trusted by your bank.
A customer gives their own AI assistant access to their bank.
Introduction, part one
Why trust in agents needs a network.
6 min 30 s
Bring us one agent, one resource and one trust boundary.
Ninety minutes. Architecture led. Your estate.