BTS-AgentBench Converts Telemetry Logs into Replayable Agent Benchmarks
August 28, 2026
BTS-AgentBench provides a deterministic pipeline to transform raw, read-only industrial telemetry into executable multi-turn agent episodes. The method normalizes metadata into a tool store and compiles static tasks with gold answers, producing a 532-row release for testing operator-facing agent workflows.
HOW THIS AFFECTS YOU
●
builderYou can use this pipeline to build more reliable benchmarks from your existing production telemetry.
●
researcherThis method allows for reproducible evaluation of agents in complex, real-world industrial environments.