Multi-Agent Systems with CrewAI: Build Teams of AI Agents
CrewAI lets you build teams of specialised AI agents that collaborate to complete complex tasks. Here's how to design and implement them.
The most complex tasks require collaboration — not just one expert, but a team. CrewAI brings this model to AI, letting you build teams of specialised agents that work together.
The Multi-Agent Advantage
A single agent trying to research a topic, analyse data, write a report, and review it will struggle. The context window fills up. The quality degrades.
Multi-agent systems solve this by specialising:
Each agent has a clear role, its own tools, and focused context.
CrewAI Basics
```python
from crewai import Agent, Task, Crew
researcher = Agent(
role="Senior Research Analyst",
goal="Find comprehensive information on the given topic",
backstory="Expert researcher with 10 years of experience",
tools=[search_tool, scrape_tool],
llm="claude-sonnet-4-6",
verbose=True,
)
writer = Agent(
role="Content Writer",
goal="Write clear, engaging content based on research",
backstory="Professional writer specialising in technical topics",
llm="claude-sonnet-4-6",
)
research_task = Task(
description="Research the latest developments in {topic}",
agent=researcher,
expected_output="A detailed research report with key findings",
)
writing_task = Task(
description="Write a blog post based on the research",
agent=writer,
context=[research_task],
expected_output="A 1000-word blog post",
)
crew = Crew(
agents=[researcher, writer],
tasks=[research_task, writing_task],
verbose=True,
)
result = crew.kickoff(inputs={"topic": "AI implementation in healthcare"})
```
Process Types
Sequential — tasks run one after another (default)
Hierarchical — a manager agent delegates to workers and reviews results
```python
from crewai import Process
crew = Crew(
agents=[manager, researcher, writer],
tasks=[task1, task2, task3],
process=Process.hierarchical,
manager_llm="claude-opus-4-6",
)
```
Designing Good Agents
The quality of your multi-agent system depends on agent design:
Clear roles — each agent should have a single, well-defined responsibility
Specific backstories — detailed backstories improve agent behaviour
Right tools — give agents only the tools they need for their role
Appropriate LLMs — use powerful models for reasoning, cheaper ones for simple tasks
Real-World Use Cases
Multi-agent systems are complex to build well. Book a call to discuss your use case.
Ready to implement AI in your business?
Book a free 30-minute strategy call — no commitment required.