IBM and UC Berkeley Diagnose Why Enterprise Agents Fail Using IT-Bench and MAST

1. 1Use IT-Bench to evaluate the performance of enterprise agents
2. 2Analyze the results to identify failure modes and areas for improvement
3. 3Apply MAST to diagnose and address the root causes of failure
4. 4Compare the performance of different models, such as Gemini-3-Flash and Kimi-K2, to identify best practices

AI researchers and engineers on a team can benefit from understanding the limitations of enterprise agents, and how to diagnose and improve their performance using tools like IT-Bench and MAST.

💡 Enterprise agents can fail due to various reasons, and using tools like IT-Bench and MAST can help diagnose and improve their performance

💡 IBM & UC Berkeley diagnose enterprise agent failures using IT-Bench & MAST