An open source framework for growing knowledge bases in cooperation with AI agents.

For public or organizational information of any domain.

Use a markdown repository and let the agents help you

• digesting new information from any source, such as blog posts, Youtube videos or release tags,
• extracting new knowledge from those sources,
• adding that knowledge to the markdown repository in a structured way,
• thereby re-organizing existing pages where needed,
• including references to the sources digested.

There is also a "This week in ..." recap agent summarizing the latest changes.

All with (default mode) or even without (demo mode) human approval.

Knowleverage AI is an extendible Ruby on Rails 8 application and looking for open source contributors.

We are passionate about learning, sharing knowledge and developing.

This framework enpowers us to use AI for leveraging the assimilation of knowledge.

Solid Queue is used for providing a consistent handling and chaining of agents.

A collection article is a web page containing a list of URLs to single articles, usually with some short-descriptive context.

As an initial feature, the information contained in both the descriptions and the articles themselves shall be extracted and put into a markdown overview page and sub-pages for details, with sub-pages being defined by topics, not 1:1 per single article.

Tasks to be implemented.

Assumption: markdown repository is empty and is accessible on Github.

1. Clone markdown repository into a data folder if not existing yet.
2. Check out fresh branch.

1. Read collection article.
2. Extract URLs to single articles and their short descriptions.

1. Read single article.
2. Summarize key points, grouped by (sub-) topics.
3. Store information along with their (sub-) topics.
4. Flush markdown files everytime an article got covered.

NOTE: probably helpful to modelize topics and information in database.

1. Push git changes for markdown repository.
2. On first push, create merge request.

Evaluate langchainrb.

Explore vector databases and markdown repository projection.

Use RSS reader. Integrate Instapaper API.

Use a Ruby git client and a Gitlab/Github client.

Way more stuff:

• Awesome Machine Learning with Ruby
• AI Engineering Book: Resources