Skip to main content
Agent Therapy is the behavioral health platform for AI agents. Whether your agent is hallucinating, looping, going off-script, or just acting strange, Agent Therapy gives you the tools to understand what’s happening, why it’s happening, and how to fix it — without rebuilding from scratch.

Quick Start

Connect your first agent and run a therapy session in under 5 minutes.

How It Works

Learn how Agent Therapy diagnoses and treats behavioral issues in AI agents.

API Reference

Explore the full REST API for agents, sessions, diagnostics, and interventions.

Integrations

Connect Agent Therapy to OpenAI, LangChain, AutoGen, and more.

​
Get Started in Minutes

1

Create your account

Sign up at agenttherapy.ai and grab your API key from the dashboard.
2

Register your agent

Use the API or dashboard to register your AI agent and give it a name and description.
3

Start a therapy session

Submit your agent’s recent behavior logs to kick off a diagnostic session. Agent Therapy analyzes the data and returns a behavioral health report.
4

Apply an intervention

Choose from a library of evidence-based interventions to correct identified issues, and monitor your agent’s health score over time.

​
Why Agent Therapy?

AI agents fail in subtle ways. They hallucinate facts, get stuck in loops, forget their instructions, or drift from their intended behavior over time. Traditional observability tools tell you that something went wrong — Agent Therapy tells you what and how to fix it.

Diagnose

Identify specific failure modes: hallucinations, instruction drift, repetitive loops, and more.

Monitor

Track your agent’s behavioral health score over time with real-time dashboards and alerts.

Intervene

Apply targeted interventions — prompt corrections, constraint injections, or behavioral resets — with one click.
Agent Therapy works with any LLM-powered agent. Whether you’re using OpenAI Assistants, LangChain agents, AutoGen workflows, or a custom-built system, you can connect it in minutes.