crewaiinc/crewai

CrewAI is a Python framework for building systems where multiple AI agents work together to complete complex tasks. Think of it like assembling a team of specialists: one agent might be responsible for research, another for writing, and another for reviewing, and CrewAI handles the coordination between them so that each agent's output feeds into the next agent's work without you having to manually orchestrate every step. The framework offers two main building blocks. Crews let you define agents with specific roles, goals, and backstories, assign them tools like web search or code execution, and set up tasks for them to complete collaboratively. Flows give you event-driven control over multi-step processes, letting you chain together individual AI calls or entire crews in a precisely defined sequence, which is more suitable for production workflows where you need predictable, auditable execution paths. The framework is built from scratch in Python and does not depend on LangChain or other agent frameworks, which the authors position as an advantage for performance and customization. It connects to language models from OpenAI, Anthropic, and many others through a unified interface. You would use CrewAI when a task is too complex or broad for a single AI call but follows a pattern that can be broken down into specialist roles, such as researching a topic and then writing a report, analyzing a codebase and proposing improvements, or running a multi-step data pipeline. It is aimed at developers and enterprises building AI-driven automation. The tech stack is Python, with configuration often done in YAML files that define agents and tasks, and the package is installable via pip.