Voice / Call Insights
How each agent call meets your business systems
Call Insights shows whether the agent did useful work — skills that hit your APIs, knowledge-base answers that resolved the question, and outcomes you can report — not only latency numbers.
- Voice OrchestratorVAD, barge-in, ASR / LLM / TTS pipeline
- Telephony GatewaySIP trunking, agent routing & human escalate
- GuardrailsVerification, HITL & PII controls
Business outcomes first
Buyers and ops leads care whether the agent helped a customer finish something. Call Insights aggregates what happened on the call against your systems:
- Skills / tools — how many business API calls succeeded, failed, or were skipped (orders, bookings, tickets, CRM)
- Knowledge answers — how many turns were grounded in your KB vs model-only replies
- Containment & handoff — resolved in-agent vs escalated to a human, with reason when available
- Scenario outcomes — return authorized, reservation confirmed, ticket routed — tied to the skill that finished the job
Why this reflects AI helping humans
Latency alone does not prove value. A fast wrong answer still burns a human. Skill success rates and KB hit rates show where the agent took work off your team — and where it still needs a person or better tooling.
Performance under the hood
For platform and SRE teams, the same session stream still includes turn timing, RTT, and TTFT so you can prove voice quality under load. Those metrics support the business story; they are not the story.
← Voice AI Agent·Open the Call Inspector on the Voice concierge to see knowledge answers per turn.