# Turn a complex AI workflow into an open-source demo

> google-research/paper-orchestra · GitHub · https://github.com/google-research/paper-orchestra
> Daily new stars: – → 1.5K (2 trending day(s) in our data: too few for a baseline, cold-start)
> Grade B (50/100, Can try, clear weak spots) · direction: Developer tools & APIs · mode: agent

## Why it moved
- [fact] The repository provides a multi-agent framework for automated AI research paper writing. (https://github.com/google-research/paper-orchestra)
- [inferred] Documentation may have helped developers understand and adopt the framework. (https://docs.tinyfish.ai/cli)
- [not_found] Trigger not found (https://github.com/google-research/paper-orchestra)

## Discount the numbers
- We have 2 day(s) of our own history for this product; the move is measured against the source's own change figure, not our baseline yet.
- Stars can be bought, and Trending brings its own stars: check forks, issues and real usage.
- Only two days of product history are available, so the result is measured against the source's change figure rather than an established baseline.
- Stars can be bought, and Trending can add stars; forks, issues, and actual usage should be checked.
- The demand-echo threads are about self-hosting and scheduled tasks rather than direct evidence of demand for PaperOrchestra.
- The supplied documentation URL does not identify itself as PaperOrchestra documentation, so its role in growth is inferred.

## Demand Echo
- 13 buyer-intent threads in 14 days
- Self-hosting Git with Forgejo while benefiting from GitHub's community — how do you manage both? (https://www.reddit.com/r/selfhosted/comments/1xxtro7/)
- Self Hosted Music Stack (https://www.reddit.com/r/selfhosted/comments/1xxhxd7/)
- EasyArr – a native Mac / iPhone / iPad / Apple TV app to manage Radarr, Sonarr, Prowlarr, SABnzbd, Seerr and Jellyfin (public TestFlight beta) (https://www.reddit.com/r/selfhosted/comments/1xxrqof/)
- AI with scheduled tasks (https://www.reddit.com/r/selfhosted/comments/1xwzkir/)

## How to follow
- play: Open-source a complex, timely AI workflow and document how to run and extend it.
- wedge: Add a hosted version, graphical interface, or vertical template built on the framework within the first two weeks.
- avoid: Do not treat stars alone as proof of retention; validate with forks, issues, runs, and repeat usage.
