What is Roark
Roark is a QA and observability platform designed for reliable Voice AI agents. It provides tools to monitor live calls, run simulations at scale, and convert failures into tests, aiming to help teams ship dependable voice agents.
Features of Roark
- Monitoring & Evaluation: Track over 40 built-in call metrics, analyze multi-speaker conversations, and run evaluators on demand or automatically.
- 40+ Built-in Metrics & Events: Monitor latency, instruction-following, repetition detection, sentiment, and custom metrics.
- Multi-Speaker Analysis: Supports calls with up to 15 speakers, offering automatic speaker identification and analysis, including talk time distribution and emotion/vocal cue detection.
- Evaluators: Run comprehensive evaluations via UI or SDK/API, with features like compliance checks (e.g., HIPAA), instruction adherence, and tool call analysis.
- Dashboards, Reports & Alerts: Build custom dashboards, schedule reports, set up alerts, and trigger webhooks for team notifications.
- Simulations & Testing: Create and run end-to-end simulations to stress-test voice agents.
- Graph-Based Conversations: Define tests as conversation flows using a graph editor, allowing for branching into edge cases and testing variants.
- Configurable Personas: Customize personas with attributes like gender, language, accent, background noise, speech patterns, emotion, and intent clarity.
- Test Generation: Automatically generate test cases from live calls.
- Integrations: Offers one-click integrations with platforms like VAPI, Retell, LiveKit, and Pipecat Cloud, as well as SDKs for Node and Python.
Pricing
Roark offers tiered pricing based on monthly call minutes and included credits:
- Startup: $500/month, includes 50,000 free monthly credits. Best for teams with up to 5,000 call minutes monthly.
- Growth: $1,200/month, includes 120,000 free monthly credits. Best for organizations with up to 12,000 call minutes monthly.
- Enterprise: Custom solution with custom credit packages, volume discounts, dedicated support, and optional on-premise deployment.
Credit costs vary per feature: Live call observability (10 credits/min), Simulations (15 credits/min), Custom metrics/events/evaluators (2 credits/call for Startup, 1.5 credits/call for Growth).
FAQ
- What does Roark do? Roark provides QA and observability for Voice AI agents, enabling monitoring, simulation, and testing.
- How difficult is it to integrate Roark? Integrations are described as one-click and taking less than 60 seconds for supported platforms.
- How do simulations work? Simulations are run as end-to-end tests using graph-based conversation flows and configurable personas.
- What are metrics and evaluators? Metrics are quantifiable data points from calls (e.g., latency, sentiment), while evaluators are tests to check for specific criteria like compliance or instruction adherence.
- What kind of sentiment analysis does Roark offer? Roark offers sentiment, emotion, and vocal cue detection as part of its monitoring features.
