This simple example creates a single agent (Writer) which receives a tech topic and creates a small blog-post. The output is a markdown file. The important thing here is the integration with ollama. Ollama allows easy access to a list of LLMs.

• Install Ollama. You can find information at Ollama or Ollama_github

• Download one model with the ollama pull {name_of_the_model} e.g. ollama pull gemma:2b to download and use locally the gemma:2b model

• Create a Modelfile. Authors of CrewAI suggest you create a custom modelfile. So, create a file named customModelfile. Inside write:

  • FROM gemma:2b
    PARAMETER temperature 0.8
    PARAMETER stop Result
    

  • ollama create CrewAI -f ./customModelfile note: Replace ./customModelfile with your actual path.

• Make sure that the model exist by showing all the models you have available with the following command:

  • ollama list

• Make sure that ollama is running with the following command:

  • ollama run CrewAI
  • To leave the chat write on the terminal /bye

• Use the command in your terminal: python -m venv /path/to/new/virtual/environment and then activate the venv.
• python -m venv .venv
• Windows version: venv\Scripts\activate
• Then install all depedencies by: pip3 install -r requirements.txt
• Then create a new folder inside your working folder e.g. crew.
• Then paste the following files inside:
  • .env
  • main.py
• Create a new folder inside and name it : blog-posts

• The only thing you need to change is the OPENAI_MODEL_NAME=. Make sure that you put the name you gave in the custom model. e.g. CrewAI

• cd into the Crew folder: cd crew
• python main.py