CrewAI is an open-source framework for orchestrating autonomous AI agents into collaborative teams. Its defining metaphor is the crew: a group of specialized agents, each given a role, a goal, and a set of tools, that work together on tasks delegated to them. Agents operate through role-playing, much as in CAMEL and MetaGPT, so a researcher agent and a writer agent might cooperate to produce a report.
The framework pairs crews with flows, which form the procedural backbone of an application. A flow manages state, drives event-driven execution, and handles conditional logic and branching - it triggers processes, delegates complex work to a crew, waits for the result, and continues. This split is meant to balance autonomy with control: crews supply agent-driven problem solving while flows supply structured, predictable orchestration around them. According to its own documentation, CrewAI is built as an independent framework rather than as a wrapper on top of another agent library.
CrewAI became one of the most popular frameworks for building multi-agent applications, frequently named alongside LangGraph and AutoGen. Its appeal is the directness of the team metaphor - describing agents by their roles and goals - which lowers the barrier to assembling a working multi-agent system, while the flows layer addresses the reliability concerns that arise once such systems run in production.