Features
Automated adversarial scenario generation
Realistic call simulation with diverse personas
Parallel voice evaluation across empathy, responsiveness, hallucinations, compliance
Production call monitoring with real-time alerting
Voice-specific quality detection (gibberish, interruptions, latency, sentiment, pitch)
Insights for root-cause clustering of LLM-judge call failures
OpenTelemetry tracing for voice agent execution
Evaluator and metric versioning
Optimize Agent button for prompt improvements via evaluators
EU region deployment for data residency
LLM judge tuning with Labs environment
Conversation analytics and flow analysis
Customizable dashboards and PDF report export
Webhook and API access, CLI and SDK
Self-improve agent via evaluator-led prompt optimization
Trace visibility for agent steps (prompts, retrievals, tool calls, outputs)
LLM-as-judge evaluation for relevance, toxicity, quality scoring
Dataset creation from traces for reproducible testing
Experiment management and regression benchmarking
Built-in Prompt IDE for iterative prompt optimization
Self-hosted deployment on local, Docker, Kubernetes
Phoenix Cloud managed hosting option
Vendor-agnostic support for any model/framework
Native OpenTelemetry integration
OpenInference specification for LLM telemetry
Human annotation and automated labeling
Ghost trajectories to simulate alternative agent paths
Eval-as-you-test for early quality feedback
One-click integration with LlamaIndex
