Junfei He

Typed provenance can prevent persistent AI agents from incorporating untrusted inputs into their autobiographical state, safeguarding user commitments and agent integrity.

Traditional logging methods waste context for AI agents, but agent-native telemetry slashes data overhead and fortifies security against adversarial threats.

The shift from deterministic to post-deterministic models reveals that classical distributed computing is just a narrow subset of a much richer landscape of participant interactions.