I did have it in mind to submit to the competition but didn’t get there in time, but I’m proud of the app so sharing it anyway:
Its called Emissary, and the idea is you make lots of focused AI assistants that can help facilitate discussion around documents, supported by Gemini’s multi-modal inputs and huge context windows.
For example, on the home page is an emissary that explains what the app does (meta), or you log in and create your own apps such as this one: https://em.sunholo.com/cUqFQeXfpoNxqCSoPFZWPIGPSRp2/a958fd22-c0c3-4b85-9542-e334ee7d9cbd
I want to eventually make lots of personas and tools that users can configure, and give it lots of configurable output options.
I like the UX of having personal, shareable bots with different personalities. I may make a few Christmas greeting elf Emissaries, but I hope to see lots of other creative ideas.
The tech stack is React, Python, Firebase, Cloud Run, Vertex, Gemini Flash, Langfuse and I’ll open-source the repo soon.