VANCOUVER, August 27, 2026 — Epistema today announced a live agent-to-agent (A2A) semantic-assurance proof exercised through Microsoft Azure AI Foundry. The demonstration marks a practical step toward portable semantic assurance across agent platforms and protocols.
In the demonstration, an upstream agent discovers Epistema through an Agent Card, invokes it via standards-based HTTPS JSON-RPC, submits a proposed commitment with evidence, and receives a structured, read-only semantic assessment. When presented with fabricated approval evidence, the system deferred rather than accepted the commitment. The readout did not authorize action or mutate governed state.
“Agent platforms provide identity, context, orchestration, tools, and execution. Epistema adds a complementary assurance layer that answers a different question: Is a consequential action warranted by the evidence and authority available now?”
— Yves Le Goff, Founder, Epistema
The release is available at www.epistema.net/a2a-release.html. The technical foundation and active architecture are available for further review.
About Epistema
Epistema is building AI workflow intelligence for governance, observability, and human–LLM collaboration. The company focuses on making semantic assurance portable across agent platforms and protocols so consequential actions can be evaluated before they are executed.
More information is available at www.epistema.net.
Media contact
Yves Le Goff
Founder, Epistema
sales@epistema.net
###