Improve discoverability on AI repositories docs

During the work on AI, we created a few different repos and services, most of them containing python code. These can be part of our infra (workflow service, AI gateway), or support (https://gitlab.com/gitlab-org/ai-powered/eli5). Accompanying these repos, we added documentation both for users of such tools but also for potential contributors (colleagues or community).

Adding documentation within this repos makes it easier to find the docs, but it makes discoverability much harder: we are often dealing with multiple of these systems at the same time, and it becomes difficult to find where each documentation is. Search is essentially impossible unless you pull the code.. Additionally, these docs are not reviewed by technical writers

doc sources

Potential solutions

  1. Move docs to gitlab docs:

docs.gitlab.com is our docs system, and that's where users and contributors expect to find all documentation. Moving the docs to gitlab would make it searchable, and we wouldn't need to have a new tool. It also means the documentation of these services matches a gitlab version, which is important for self-hosted customers.

  1. Make each repo doc searchable on gitlab docs:

We don't need to remove the docs necessarily to gitlab repo, maybe we could change docs.gitlab.com to accept multiple repos. Advantage is that we don't need to change current docs. Downside is that customers would not have these docs available.

  1. Add a feature to gitlab that allows doc search across multiple repositories:

Much more complicated, but wouldn't need us to change the docs, they would still be kept by the code, and this is something our customers might find useful as well.

Edited by Eduardo Bonet