CrewAI Basics

CrewAI is a Python framework for orchestrating role-playing, autonomous AI agents that work together like a real team

Agents are defined by three core attributes: role (job title), goal (objective), and backstory (expertise and personality)

Tasks specify what needs to be done with a description and expected_output, and are assigned to specific agents

Crews combine agents and tasks together, handling orchestration and execution flow automatically

Sequential process executes tasks one at a time in order - simple, predictable, and most common for linear workflows

Hierarchical process creates an automatic manager agent that delegates tasks and coordinates workers dynamically

Role-based design in CrewAI is more intuitive than manual agent coordination - think in terms of job roles, not code patterns

Use allow_delegation=True for agents that should be able to ask others for help (e.g., managers)

Task context parameter lets you reference outputs from previous tasks, enabling information flow between agents

Tools should match agent roles - researchers get search tools, writers get file tools, analysts get validation tools

Start with 2-3 agents and sequential process before adding complexity - test iteratively with small tasks first

Enable verbose=True during development to see detailed logs of agent thinking and tool usage for debugging

CrewAI is ideal for content creation, research, analysis - workflows where tasks naturally map to specialized roles

Memory feature (when enabled) allows agents to remember previous interactions across multiple crew runs

Always define specific expected_output for tasks to guide agents toward producing the right format and content