Claims you can check, not metrics we made up.
A trust-infrastructure company should hold itself to its own standard. So this page carries only what's actually running and independently checkable, no invented benchmarks, no test counts we won't stand behind, no customer names we can't clear.
Running today
The foundation runs on Cloud Run
SYNTHERA is not a slide. The foundation runs live, its policy engine loads rules at boot, and a request against a provisioned agent is refused or permitted by a named rule. The substrate is what stays deployed; the agent it enforces on is woken on request.
Identity proven across ADK and OpenAI
One identity model and one audit-of-record, not one per framework: a dated conformance run exercises the same signing and verification path across the ADK and OpenAI adapters, 7/7 and 9/9 respectively, in the Phase 1E run. Those are the two frameworks with a governed-agent template, and they are the two the run covered. vaid-langchain is published and installable and is a request-signing adapter, which is a smaller and different claim than governing a LangChain agent; it was not in this run. That run proves identity. Policy enforcement is proven separately, on the root agent SYNTHERA provisions, which is woken on request rather than kept running.
Verify a signed action yourself, right now
The live demo runs the real proof-of-possession path in your browser against the standard's frozen conformance vector. Tamper with the request and watch it reject. That's proof you run, not proof we assert.
Three languages, one set of bytes
The Rust, Python and TypeScript reference signers reproduce the same digest and the same signature from the same inputs, with no shared runtime, the interoperability guarantee made concrete rather than claimed.
The agent stack is hibernated between walkthroughs
The governed agents behind the public demo and behind provisioning are deliberately not left running: they are torn down between walkthroughs and brought back by a documented wake sequence, because a sandbox that bills continuously to prove a point nobody is watching is the wrong thing to pay for. What stays up is the substrate, the policy engine and the audit-of-record. What waits for a wake is the agent itself, so the live demo says so when it is asleep and offers a walkthrough instead of an error.
SYNTHERA is the trust layer for multi-agent systems: every agent gets a verifiable identity, scoped authority and a tamper-evident record, so software from different teams, vendors and frameworks can act on each other’s behalf without custom glue between every pair.
Discuss a technical evaluation.
We'll walk you through the live foundation and a verification against a framework you actually run.