Abstract:
Complete end to end irl game status screen created for motivation, using journal entry analysis.
Intended usage: record moment to moment progress (eg: attended to my microprocessors class for 50 minutes) using your voice. See instant progress against across your skills and characteristics (eg: +1 exp computer engineering) and final accumulated statistics (computer engineering total exp: 50; total level: 5, etc).
Functionality:
- journal view: voice entries and manual entries, live voice dictation / transcription, complete entry view
- graph view: custom topology to summarize a person’s soul: characteristics, skills, actions
- status screen: a simple self-status viewer to view accumulated experience.
- debugger: view all data stores.
Pipeline:
entry (voice, manual text) -> extract actions, skills, characteristics -> form topology fragment -> merge fragment with main graph -> feedforward initial time taken (30 min = 1 exp) from actions to characteristics to get exp per node -> level calculations -> update datastores -> view on status screen and entry results.
Notes:
Gemini 3 hackathon:
This was the longest hackathon I have every participated in, taking place over 2 months. I had pleasant experiences for all my past hackathons, so I thought that this wasn’t going to be any different– it completely was. Hackathons should be a maximum of 48 hours. Any longer, and it just becomes exhausting:
hackathons are a way to choose building and deployment over learning, and such a mentality is only feasible in the very short term.
None the less, I was still able to create something that I would never have been able to in such a short timeframe: a full end-to-end live-hosted ai app (Try it out, you can sign in as a guest!). I learned so so much, even though I was working alone, mainly about ai workflows itself. I didn’t know anything about react coming in, and now I’m pretty confident on using it to make a simple web page.
Submission materials (more information):
- Hosting: (https://self-statistics-system-v2.web.app/app/journal)[https://self-statistics-system-v2.web.app/app/journal] (updated: v1 -> v2 to reduce load)
- Demo video: (https://youtu.be/zvvyPhYOxqU)[https://youtu.be/zvvyPhYOxqU]
- Devpost: (https://devpost.com/software/incremental-system-irl-status-screen)[https://devpost.com/software/incremental-system-irl-status-screen]
Takeaways:
The biggest takeaway would probably be understanding the differences in responsibility between frontend (please, for your own sake, have frontend only present) and backend. Google AI Studio does only frontend, which made refactoring very very… ‘fun’.
I had also never made a full backend hosted app before this! This was my first experience using documents and NoSQL, and it went over pretty alright!
This app will eventually have to migrate away from firebase, but firebase acts as a pretty good prototyping tool for now.
V2 coming out soon! (github and final results page link to v2 because v1 was hackathon submission and is thus frozen)
Technical Details
View on github. Frontend: React Backend: Firebase tech