CRM-Agent-Bench: Benchmarking Tool-Using Agents in Stateful Enterprise Environments
4.60T1 sourcearXiv cs.HC
Source record
Published by arXiv cs.HC (T1 source). The original is at https://arxiv.org/abs/2610.32004.
Pipeline notes
The summary and note below are generated by the signal pipeline — they are Beyond Desk’s reading, not quotations from the source.
SummaryCRM-Agent-Bench introduces 555 stateful, multi-turn CRM tasks across nine business domains, with executable contracts checking actions, arguments, ordering, responses, target state, protected state, and policy compliance. Sixteen evaluated systems show similar strict-success rates but substantial differences in cost and domain reliability.
Why it mattersProvides an auditable benchmark and public task suite for testing whether agents execute consequential business workflows correctly, not merely return plausible answers.
Cited by
No citations on record.