This project is an evolving fork of the Binxly/ai-researcher repository, which itself is a fork of the original ai-researcher repository created by @mattshumer. The original version utilized Claude 3 and SERPAPI to conduct research, while the Binxly/ai-researcher fork modified the project to use Google Search API for searching capabilities. This fork builds upon those changes with further enhancements. Below are the details of the modifications:

1. Switched to OpenAI: The underlying AI model has been switched from Claude 3 to OpenAI's GPT-3 model for improved performance and flexibility.

2. Parsing Improvements: The web search function has been enhanced to parse only the relevant content from the google search result, rather returning the whole Google meta data. This change reduces the number of tokens and improves the quality of the generated content. However, the content is still not parsed from the original website. This is still a mayor drawback of the current implementation. Until this is fixed, the quality of the generated content will be limited.

Until the parsing of the content from the original website is implemented, I suggest to use the following project: gpt-researcher.

• Switch to OpenAI
• Improve parsing of the google search result to reduce amount of tokens
• Parse the content from the original website, not just from the Google meta data
• Add Azure OpenAI support

Credit goes to the original creator, @mattshumer, for the foundational work on the ai-researcher, and to Binxly for their enhancements in the Binxly/ai-researcher fork. The changes made in this fork aim to further improve upon their work.

This document outlines the changes from the joriskalz/ai-researcher fork, which itself modified the original project. It should be noted that while the core functionality remains the same, additional optimizations and features have been introduced to enhance the AI Researcher tool.

This document is for the forked project and should be used in conjunction with the documentation from the Binxly/ai-researcher fork and the original project for a complete understanding of the system's capabilities and usage.