τ³-Bench Banking
Fintech customer-support benchmark for knowledge retrieval plus tool-using conversational agents.
τ³-Bench Banking is the banking/fintech customer-support benchmark domain in Sierra's τ³-Bench / τ-Knowledge work. It evaluates language-agent performance on knowledge-grounded multi-turn banking workflows that require searching an unstructured policy corpus and executing multi-step tool calls, with success graded by backend database state.
Recent stories
0 linked stories
No linked stories yet.