RELAI SDK

RELAI SDK is an open-source framework that builds reliable AI agents through a continuous loop of simulation, evaluation, and optimization.

RELAI SDK hardens agentic workflows by replacing guesswork with a structured loop: simulate, evaluate, and optimize. Use the toolkit to stress-test agents against LLM personas, run automated benchmarks with code-based judges, and leverage Maestro (the platform's holistic optimizer) to tune prompts or suggest graph-level changes. It plugs into existing frameworks (including LangGraph and OpenAI) to turn execution traces into actionable improvements, ensuring your agents perform predictably in complex environments.

https://relai.ai