I built a similar product called Automagical awhile back (https://automagical.ai/) that we exited to a media agency.
A few notes from my experience:
1. Using gifs in this way as part of a composite video is unfortunately against giphy's ToS. We received a cease & desist from them after awhile, so just be aware of that up front.
2. If you want to support more variety in your transitions, check out my OSS project https://github.com/transitive-bullshit/ffmpeg-concat
Overall, the product looks really polished & I love where you're headed. Best of luck!