05/31/2026
Benchmark scores don't tell you if your agent will hold up in production. If you've built eval infrastructure that successfully predicts real-world behavior or begun a deep exploration in this area, then AGNTCon + MCPCon North America wants to hear from you.
The Evaluation and Testing track is looking for practitioners working on the hard parts: ground truth design at scale, regression coverage across non-deterministic workflows, LLM-as-judge calibration, and closing the gap between eval pass rates and real task success.
CFP closes in ONE WEEK on June 7. Take the stage October 22–23, San Jose. 🎤 Submit a proposal: https://bit.ly/4tFfCdz
🎟️ Attending? This ticket is heating up! Register now and save up to $550: https://bit.ly/4vkntyD