UniPat AI team debuts SaaS-Bench as agents finish under 4% of real SaaS workflows

Built across 23 live SaaS apps and 106 long-horizon tasks, the open benchmark finds frontier agents stumble on planning, memory, cross-app context, and error recovery.

By · Published

Illustration of a UniPat AI agent struggling to complete a complex multi-application digital workflow for the SaaS-Bench benchmark.

Built across 23 live SaaS apps and 106 long-horizon tasks, the open benchmark finds frontier agents stumble on planning, memory, cross-app context, and error recovery.