What is CrewAI?
CrewAI is a Python framework for building multi-agent AI systems. You define agents with specific roles ("Researcher", "Writer", "Analyst"), give them goals and tools, and then let them collaborate on tasks. Agents can work sequentially or in parallel, delegate subtasks to each other, and share context.
This is useful for complex workflows where a single LLM call isn't enough: research tasks that need multiple steps, content pipelines that involve drafting and editing, or analysis workflows that combine data gathering with synthesis. CrewAI uses LangChain's ChatOpenAI under the hood, so pointing it at EUrouter is the same one-line change. Every agent call, regardless of which agent makes it, goes through EU infrastructure.
Quick to integrate
A few lines of code is all it takes. Swap your base URL and you are routed through EUrouter.
1from crewai import Agent, Crew, Task
2from langchain_openai import ChatOpenAI
3
4llm = ChatOpenAI(
5 model="gpt-4o",
6 api_key="eur-...",
7 base_url="https://api.eurouter.ai/v1",
8)
9
10researcher = Agent(
11 role="Researcher",
12 goal="Find key information",
13 llm=llm,
14)Get started in minutes
Follow these steps to connect your application to EUrouter.
Install CrewAI
Install CrewAI and the LangChain OpenAI integration.
1pip install crewai langchain-openaiConfigure the LLM
Create a ChatOpenAI instance pointing to EUrouter.
1from langchain_openai import ChatOpenAI
2
3llm = ChatOpenAI(
4 model="gpt-4o",
5 api_key="eur-...",
6 base_url="https://api.eurouter.ai/v1",
7)Build your crew
Create agents and tasks as usual. All LLM calls route through EU infrastructure.
1from crewai import Agent, Crew, Task
2
3researcher = Agent(
4 role="Researcher",
5 goal="Find key information",
6 llm=llm,
7)
8
9task = Task(
10 description="Research EU AI regulations",
11 agent=researcher,
12)
13
14crew = Crew(agents=[researcher], tasks=[task])
15result = crew.kickoff()Why use CrewAI with EUrouter
Multi-agent, EU-compliant
A CrewAI crew with five agents running ten tasks might make dozens of LLM calls. Each of those calls sends prompts and receives completions. With EUrouter, every single one of those calls stays within European infrastructure, regardless of how complex your workflow gets.
- Every agent call routed through EU servers, including delegation between agents
- Sequential and parallel task execution both route through EUrouter
- Inter-agent communication context stays GDPR-compliant throughout the workflow
- EUrouter's dashboard gives full visibility into which agent used which model and when
Mixed-model crews
Different agents need different models. A researcher might need a fast, cheap model for scanning lots of sources. An analyst might need GPT-4o for complex reasoning. A writer might work best with Claude. With EUrouter, you assign different models to different agents using a single API key.
- Assign the right model to each agent role based on the task requirements
- Optimize costs by using cheaper models for high-volume, simpler agent tasks
- EUrouter's smart routing can also automatically select the best provider
- One API key for all agents, all models, all providers
Budget control for agent workflows
Multi-agent workflows can run up costs quickly, especially with long-running tasks or autonomous agents. EUrouter lets you set spend limits per API key. If a crew hits its budget, it stops rather than running up an unlimited bill.
- Per-key spend caps prevent runaway costs from autonomous agent workflows
- Real-time cost tracking in the EUrouter dashboard, broken down by model
- Set different budgets for development, staging, and production crews
- See exactly which agents and tasks are responsible for your model costs