Product · Post-call analysis

Every production call, scored and explained.

Roark analyzes every call as it lands: 64+ audio-native metrics, issues filed automatically, alerts on any threshold, traces and dashboards on tap. For voice calls and chat threads alike.

Get started free Book a demo

Start free with $50 in credit, no card needed. Works with Vapi, Retell, LiveKit, Pipecat + your stack.

Scoring production voice AI for teams at

radiantgraph

01 · Post-call analysis

64+ metrics. Your models,
not just an LLM.

Every production call scored as it lands: issues filed, alerts fired, dashboards and OTEL traces on tap, for voice calls and chat threads alike. And where most tools grade a transcript with an LLM, Roark runs purpose-built audio models on the call itself, measuring what your customer actually heard.

platform.roark.ai/calls/c_8f42

+1 (415) 555-0134 → support_v2

Today 14:32 · 3m 42s · Vapi

Caller: I was told the refund would arrive by Friday…

Agent: Let me check that for you, one moment.

Dead air: 3.8s before the agent responded.Issue #482 filed

Metrics

Pronunciation

Empathy

Instruction following

Response time

Disclosures ✓Identity check ✓+ 60 more scored

Audio-native

custom models

Pronunciation
Accent clarity
Emotion
Vocal stress
Pace & pauses
Interruptions

Conversational

LLM + rules

Resolution
Empathy
Task success
Hallucination
Repetition
Tone

Compliance

policy

Disclosures
PII exposure
Identity check
Script adherence

Performance

latency

Time-to-first-word
Turn latency
ASR WER
Barge-in handling

64+metrics out of the box

∞custom metrics, your rules

Audio + LLMmodels on every call

See a scored call on your own audio

02 · Beyond the score

The score is the start.
Operating is the point.

When a call goes wrong, Roark files the issue, pages your team and hands you the trace: the tooling to run voice AI like production software, not a black box.

Issue tracker

Every failure is filed automatically, clustered with its lookalikes, and tracked across deploys until the recurrence count reads zero.

Issues · support_v21 open · 2 resolved

#482Dead air after account lookup12 callsopen

#479Refund issued before identity check31 callsfixed

#463Drug names mispronounced8 callswatching

Alerts & monitors

Put a threshold on any metric. The moment one crosses, Slack or your webhook hears about it, with the failing calls attached.

Monitor · resolution ratethreshold 85

Resolution rate · last hour91→78

Crossed at 14:32: #voice-alerts and the on-call webhook pinged, with the 6 failing calls attached.

Tracing

Every turn broken into spans across ASR, LLM, TTS and network, so a slow answer points at the component that caused it.

Trace · turn 7 of 221.65s total

ASR

320ms

LLM

980ms

TTS

210ms

Network

140ms

the dead air lives in the LLM span · OTEL-compatible, exportable

Dashboards & reports

Score trends by agent, version, metric and language. When someone asks whether the fix worked, the chart is already made.

Dashboard · quality scoreby agent version

8 weeksv4 shipped ↑

Quality since the v4 deploy71→94

03 · Get started

First call scored in under a minute.

One click on any platform below and production calls stream in on their own, or send any recording with a few lines of code.

Read the quickstart

evaluate.ts

import Roark from '@roarkanalytics/sdk'
const client = new Roark({ bearerToken })
await client.call.create({
  recordingUrl, startedAt,
  interfaceType: 'PHONE',
  callDirection: 'INBOUND',
  agent: { customId: 'support_v2' },
}) // scored in seconds

Node · Python, plus a REST API for CI/CD and webhooks the instant a call is scored

Works with