Skip to main content
OneRun is an open-source platform for evaluating AI agent performance through realistic conversation simulations. Deploy it yourself, customize it for your needs, and build better agents with complete control over your evaluation process.

What OneRun Does

  • Test your agents against diverse, AI-generated personas in realistic scenarios
  • Measure performance with custom objectives and scoring criteria
  • Track improvements over time as you iterate on your agent
  • Deploy anywhere - self-hosted, on-premise, or in your preferred cloud

How OneRun Works

1

Define Your Agent

Create an agent profile that represents your AI system and its intended purpose
2

Set Evaluation Objectives

Define what success looks like with custom scoring criteria for different aspects of performance
3

Create Test Scenarios

Design realistic simulation scenarios that reflect your agent’s real-world use cases
4

Generate Diverse Personas

Automatically create varied conversation participants with unique backgrounds and characteristics
5

Run Conversations

Execute realistic dialogues between your agent and generated personas through your custom worker
6

Analyze Results

Review detailed performance metrics, identify patterns, and discover improvement opportunities

Why Choose OneRun?

Open Source & Self-Hosted
Full control over your data, evaluations, and deployment. No vendor lock-in.
Framework Agnostic
Works with any AI framework - LangChain, direct API calls, or custom solutions.
Production Ready
Built for teams serious about AI quality with systematic evaluation processes.

Get Started Today

Ready to Build Better Agents?

OneRun transforms agent development from guesswork into a systematic, data-driven process. Since it’s open source, you have complete control over your evaluation process and data.
OneRun is designed for teams serious about AI agent quality. Deploy it anywhere and customize it for your specific needs.