Noveum.ai
Noveum docs

Noveum documentation

Instrument, observe, evaluate, and improve production AI applications from one connected workflow.

Noveum brings tracing, dataset preparation, evaluation, synthetic testing, and recommendations into one platform. Start with the workflow closest to what you need today.

The Noveum workflow

Instrument

Add the Python SDK or use a framework guide for LangChain, LangGraph, LiveKit, or Pipecat.

Observe

Inspect request paths, latency, token usage, model inputs, outputs, and failures in the dashboard.

Prepare data

Convert selected traces into a reusable dataset. Use ETL jobs when production data needs mapping or normalization.

Evaluate

Create an Eval Job, choose relevant scorers, and review per-item reasoning with NovaEval.

Improve and test

Generate focused recommendations with NovaPilot, then validate new behavior at scale with NovaSynth.

Pick an integration

ApplicationStart hereWhat it covers
Python or a custom LLM appSDK integrationInitialization, traces, spans, and configuration
LangChainLangChain overviewChains, tools, agents, and callbacks
LangGraphLangGraph overviewNodes, transitions, and agent workflows
LiveKitLiveKit overviewVoice sessions, STT, TTS, and latency
PipecatPipecat overviewVoice pipelines and synthetic testing
CrewAICrewAI overviewMulti-agent tasks and handoffs
Claude Code, Codex, Cursor, and VS CodeMCP server referenceResources, prompts, OAuth, and API key setup

Product guides

Reference

If you are new to Noveum, continue with the five-minute quick setup.