Brian Douglas

Agents produce high-fidelity traces of everything they do. But traces alone don't produce memory. This talk introduces three primitives for turning session history into something actionable:

AI agents forget everything between sessions. They loop, burn tokens, and repeat the same failures because no individual turn has context about the ones before it. tapes is an open source telemetry proxy that intercepts LLM API calls, builds a content-addressable Merkle DAG of every conversation turn, and publishes events to Kafka with zero additional instrumentation required (no SDK). The agent points its API base URL at the proxy and keeps running. From there, Flink SQL runs continuous anomaly detection against the telemetry stream: stuck loop detection in tumbling windows, token spike detection against rolling averages, and domain-specific pattern matching. When Flink catches something, it publishes an alert to a separate Kafka topic. An alerts consumer writes that back into the agent's session store. The next run reads those signals before it starts. To test this architecture out, we did what we knew best which was built and tested this by speedrunning parallel agents evolving their navigation parameters across Gameboy games. The infrastructure around it was observant. The Merkle DAG structure means every conversation turn is content-addressable and forkable, so you can…

When developing modern APIs, there a few options to choose from. REST is not a standard  but instead style. Trust is constantly broken between the API and client. gRPC and GraphQL provide standards but require complex tooling.