Abstract
This publication discloses, in enabling detail, a mechanism that couples a synchronous conversational assistant (a chat surface a human talks to and gets an immediate answer from) to an asynchronous autonomous-agent runtime (a background "persona" that perceives, reasons, plans, acts, and reflects on its own schedule) through a single caller-trust-scoped ingestion valve. Every conversational turn is stamped with an access mode denoting the trust of the caller. The valve admits a turn as cognitive work only when the access mode denotes the resource owner; turns from untrusted (guest) callers are dropped at the valve so untrusted conversation never becomes autonomous cognition. Admitting an owner turn lazily activates the addressed persona if it is dormant — creating an in-memory process-table entry on demand instead of eagerly booting every persona — and enqueues the full turn as a human-originated attention item consumed later by the asynchronous cognitive loop. The runtime keeps an operating-system-style in-memory process table whose every lifecycle transition is written through to a durable cognitive-state row, so the live set of activated personas is rebuilt deterministically on boot after a crash or redeploy. The novelty being staked is the combination: a trust axis that decides whether conversation becomes cognition at all, trust-triggered lazy activation, and crash-restorable process state — deliberately drawn around any pre-existing per-conversation mode machine. This document is published to establish dated public prior art and to bar patenting of the disclosed mechanism.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Assuncao, gustavo matthew, "Owner-Gated Conversation-to-Cognition Ingestion with Lazy Persona Activation and Crash-Restorable Process Table", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/10930