CrewAI with SovereignEG

Goal: run one CrewAI agent on a SovereignEG model, then a two-agent crew.

Time: 10 minutes.

CrewAI talks to OpenAI-compatible servers through its LLM class. The only special rule: the model name must start with openai/ so CrewAI knows which API format to use.

Step 1 — Install

CrewAI needs Python 3.10 – 3.13.

pip install crewai

Step 2 — One agent

Save as crewai_hello.py:

import os
 
from crewai import Agent, LLM
 
llm = LLM(
    model="openai/SovereignEG/Qwen3.8-27B-FP8",   # "openai/" prefix + SovereignEG model id
    base_url="https://backend.sovereigneg.com/v1",
    api_key=os.environ["SOVEREIGNEG_API_KEY"],
)
 
agent = Agent(
    role="AI Assistant",
    goal="Help the user",
    backstory="You are a helpful AI assistant.",
    llm=llm,
)
 
result = agent.kickoff("Hello")
print(result.raw)

python crewai_hello.py

If your CrewAI version says the provider is unknown, add custom_openai=True to the LLM(...) call. Newer versions accept the openai/ prefix alone.

Step 3 — A crew of two agents

Agents get tasks; a crew runs the tasks in order and passes results along.

import os
 
from crewai import Agent, Task, Crew, LLM
 
llm = LLM(
    model="openai/SovereignEG/Qwen3.8-27B-FP8",
    base_url="https://backend.sovereigneg.com/v1",
    api_key=os.environ["SOVEREIGNEG_API_KEY"],
    temperature=0.3,
)
 
researcher = Agent(
    role="Researcher",
    goal="Collect clear facts about {topic}",
    backstory="You are careful and only state facts you are sure of.",
    llm=llm,
    verbose=True,
)
 
writer = Agent(
    role="Writer",
    goal="Turn facts into a short, simple article",
    backstory="You write in easy English for busy readers.",
    llm=llm,
    verbose=True,
)
 
research = Task(
    description="List 5 key facts about {topic}.",
    expected_output="A bullet list of 5 facts.",
    agent=researcher,
)
 
write = Task(
    description="Write a 150-word article using the facts from the research task.",
    expected_output="A short article with a title.",
    agent=writer,
    context=[research],          # the writer sees the researcher's output
)
 
crew = Crew(agents=[researcher, writer], tasks=[research, write])
result = crew.kickoff(inputs={"topic": "the Suez Canal"})
print(result.raw)

verbose=True prints every step, so you can watch the agents think and hand work to each other.

Step 4 — Give an agent a tool

from crewai.tools import tool
 
@tool("Exchange rate")
def exchange_rate(currency: str) -> str:
    """Return today's rate of 1 USD in the given currency."""
    return {"EGP": "48.5", "SAR": "3.75"}.get(currency.upper(), "unknown")
 
analyst = Agent(
    role="Analyst",
    goal="Answer money questions",
    backstory="You always use the tool for rates.",
    llm=llm,
    tools=[exchange_rate],
)
print(analyst.kickoff("How many EGP is 100 USD?").raw)

Troubleshooting

Problem	Fix
`Provider ... not found` / `BadRequestError`	The model string must be `openai/SovereignEG/Qwen3.8-27B-FP8` — keep the `openai/` prefix in front of the SovereignEG id
Slow or long runs	Set `max_iter=5` on the `Agent` and `temperature=0` on the `LLM`
Python version error	CrewAI does not support Python 3.14 yet; use 3.12

Next: LlamaIndex