Benchmark RadarAI BENCHMARK PROFILE
IHBench
IHBench evaluates post-interruption recovery in voice agents executing state-machine-driven workflows across 10 enterprise domains. It scores task fulfillment and recovery quality for six interruption types.
- Released
- 2026-06-17
- Readiness
- Paper only
- Primary field
- General AI
Why it matters
Voice agents must handle interruptions while maintaining workflow progress, but existing benchmarks measure only interruption timing. IHBench focuses on recovery quality, a distinct capability axis important for deployed agents.
Motivation
Voice agents deployed in structured workflows (customer service, healthcare scheduling, account management) must handle frequent user interruptions while maintaining progress through multi-step procedures.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.