CrewAI - AI-Powered Blogging Agents using LM Studio, Ollama, JanAI & TextGen

---

---

AI Summary

Summary: Integrating Open Source Language Models for Blog Creation

• Introduction:
  • Demonstrating the use of Crew AI to integrate multiple open-source language models.
  • Encourages subscribing to the YouTube channel for AI-related content.
• Installation Steps:
  • Install necessary packages: Lang chain Community, Crew AI, Lang chain OpenAI, and Dougduug Go Search.
  • Create app.py and import required modules from Crew AI, Lang chain, and community tools.
• Tool Initialization:
  • Initialize search tool and define language models using LM Studio, Jan AI, and Text Generation Web BUI.
  • Download and set up models, ensuring proper context length for blog post generation.
• Agent Creation:
  • Researcher Agent: Uses LM Studio and Dougduug Go Search for internet research.
  • Insight Researcher Agent: Extracts key insights from the Researcher Agent’s data.
  • Writer Agent: Writes content based on insights using Olama.
  • Format Agent: Formats text into Markdown for platforms like WordPress.
• Task Assignment:
  • Research Task: Identify the next big AI trend.
  • Insight Task: Find key insights from research data.
  • Write Task: Compose a blog post on the findings.
  • Format Task: Convert text to Markdown format.
• Execution:
  • Instantiate a Crew with agents and tasks, set the process to sequential.
  • Begin task execution with check_crew.kickoff, save, and print results.
• Running the Code:
  • Execute app.py in the terminal.
  • Monitor the process through logs and system monitor for RAM and CPU usage.
  • Adjust models and prompts as needed to improve output quality.
• Troubleshooting:
  • Address issues with model performance and integration.
  • Suggest using unquantized versions and proper prompt engineering for better results.
• Comparison with OpenAI:
  • Remove local language models and use OpenAI instead.
  • Export OpenAI API key and rerun the code.
  • Observe clearer results with OpenAI, including well-formatted Markdown content.
• Conclusion:
  • Emphasizes the importance of model choice and prompt engineering.
  • Promises more similar videos and encourages likes, shares, and subscriptions.