Train AI Agents Like Humans
Create AI agents using human-readable and LLM-executed playbooks. Simple, flexible, and powerful.
GitHubThis is a playbook
This is an AI Agent running the above playbook
Output will appear here...
Why Playbooks?
It all started with a simple question - Why can't we train AI agents just like we train human agents using training material and playbooks?
One of the biggest challenges in building AI agents today is specifying and modifying their behavior. Code is too complex for business users, UI builders lack flexibility, and LLMs struggle with complex prompts.
Playbooks is the perfect middle ground. Write agent behavior in easily readable English-like pseudocode, while the framework handles:
- Step-by-step control flow
- Internal and external tool integration
- Complex behaviors with multiple playbooks
- Multi-agent communication
- Event-triggered automation