Trane cuts HVAC diagnosis 60x with Amazon Bedrock AgentCore

Trane's starting point was a familiar one for large asset operators: millions of connected HVAC assets feeding Trane Cloud, but answers scattered across dashboards and menus. The company says a single operational question could take 20 minutes or more, and different roles—field technicians, account managers, building owners—needed radically different views of the same telemetry.

Rather than a single monolithic agent, Trane split the system into five specialized assistants governed by separate system prompts, connected to Trane Cloud APIs through AgentCore Gateway and the Model Context Protocol. The team chose AgentCore over hosting on Amazon ECS or AWS Lambda, citing managed session isolation, built-in memory, native tool orchestration, and framework-agnostic design. During development, AgentCore Observability helped trace a recurring tool failure to a missing secret in AWS Secrets Manager.

Looking ahead, the outcome hinges on whether automated accuracy thresholds from AgentCore Evaluations can reliably gate releases, and whether opening the Gateway to other Trane engineering teams leads to genuine reuse or duplicate work. For now, Trane's phased rollout—3–4 weeks to prototype, internal beta with field technicians, then external release—offers a concrete example of turning proprietary operational data into a conversational interface, though the 60x claim remains based on Trane's own benchmarking rather than outside verification.