AI-Journalist is an experimental tool that leverages the Claude 3 AI model to automate the research, writing, and editing of articles on any given topic. It is designed for users who want to explore AI-assisted content creation for journalism or general writing tasks. The project aims to produce high-quality, informative, and engaging articles by integrating web search and AI analysis.

How It Works

The system initiates by prompting the user for a topic. It then uses Claude 3 Haiku to generate relevant search terms, which are fed into the SERP API for web searches. Claude 3 is employed again to select the most pertinent URLs from the search results. The newspaper3k library extracts article text from these selected URLs, and Claude 3 is used to write an initial draft and then revise it based on editorial feedback generated by another Claude instance.

Quick Start & Requirements

• Install/Run: Open claude_journalist.ipynb in Google Colab or Jupyter Notebook.
• Prerequisites:
  • Anthropic API key
  • SERP API key
• Setup: Replace placeholder API keys in the notebook and run cells sequentially.
• Links: HyperWrite Platform (for a no-code version)

Highlighted Details

• Utilizes Claude 3 Haiku for search term generation and Claude 3 for article writing and editing.
• Integrates SERP API for web search and newspaper3k for article text extraction.
• Offers a no-code version on the HyperWrite Platform.
• Includes a disclaimer emphasizing the need for human review and fact-checking.

Maintenance & Community

• Contact: Matt Shumer (@mattshumer_)
• Suggestions for improvement include enhancing search result filtering, implementing fact-checking, and optimizing article structure.
• Users are encouraged to contribute via issues or pull requests.

Licensing & Compatibility

• License: MIT License
• Compatibility: Permissive for commercial use and integration with closed-source projects.

Limitations & Caveats

The output is experimental and requires careful human review and fact-checking before publication. Generated articles may need further editing to meet specific editorial standards. Known improvement areas suggest the current fact-checking and verification mechanisms may be basic.