RunLog AI Documentation

Everything you need to add comprehensive observability and control to your AI agents with minimal code changes. Simple decorators, flexible policies, and powerful experiments.

New to RunLog AI?
Start with our quick start guide - add observability with decorators, configure policies your way, and track experiments effortlessly.

Popular Topics

Quick Start
Get up and running with RunLog AI in under 5 minutes
Learn more
Python SDK
Complete reference for the Python SDK
Learn more
Policy Engine
Learn how to create and manage safety policies
Learn more
Replay System
Understand deterministic replay and testing
Learn more

What is RunLog AI?

RunLog AI is a minimally invasive observability and control platform for AI agents. Add decorators for automatic tracking, define policies through YAML, GUI, or natural language, and run experiments with simple parameter changes.

Flight Recorder

Complete step-by-step traces of agent execution including prompts, tool calls, costs, and latency. Add @runlog decorator to any function.

Policy Firewall

Runtime policy enforcement to prevent unsafe actions, budget overruns, and compliance violations. Define policies via YAML, GUI, or natural language.

Replay Lab

Deterministic replay system for testing changes and debugging issues without affecting production. Run experiments via decorator parameters.