Tip Of AI
Discover AI tools
Noveum is an AI agent reliability platform for evaluating production chat agents, voice agents, and multi step workflow agents. It creates a closed loop process that captures agent traces, converts selected production runs into evaluation datasets, applies compatible scorers, investigates failures, and produces ranked recommendations for improvement. The platform includes noveum trace, an open source Python SDK for recording language model calls, tool invocations, retrieval events, custom spans, and agent steps. It supports LangChain, LangGraph, LiveKit, Pipecat, CrewAI, OpenTelemetry, REST ingestion, and high volume trace collection. NovaEval provides a library of 112 published scorers across 15 categories, including 36 voice and audio scorers. NovaPilot analyzes evaluation evidence and generates reports covering health, scorer performance, recurring failure patterns, system prompt findings, confidence, and implementation effort. NovaSynth supports synthetic testing for configured voice and chat agents using personas and scenarios. NovaGuard is a beta feature for cost caps and request or token limits. Noveum offers managed cloud deployment, enterprise on premise options, in virtual private cloud deployment, and Bring Your Own ClickHouse configurations. It provides a free tier with guided setup, while enterprise features may include SSO, access controls, auditability, and deployment customization. SOC 2 Type II is in progress and GDPR requirements are under review.
