Tracker Addict App

Track anything, find patterns, and create custom journals easily with AI-powered natural language input. Tracking made simple for everyone.

Comment

DISCLAIMER

I thought that demo is a good opportunity to try ElevenLabs, plus the hackathon is all about AI generation. So I thought it would be a good idea to make demo generated with AI. So I did it but now I doubtful because I hadn't enough time to make it perfect and standout. I hope that quality of my demo won't affect negatively the project it presents, thank you for your attention!

Inspiration

Very often I want to track something in a format that I can't find in existing apps so I end up tracking things in notes. So I thought wouldn't it be convenient to have such an app that would allow me define logs by myself? Then I thought it will also allow me to do analytics based on different types of logs which specialized logging apps cannot offer. For a simple example, I could find correlation between how many excercises I did and my monthly costs on groceries.

What it does

Alright, the fact is that my app essentially acts as a form constructor. It lets user define a custom log, which is essentialy a form schema, and then make records in a format defined by the schema. But the most interesting feature of this app is AI UI which takes all the complexity of defining schemas, filling in forms, specifying values for dictionaries on itself. User just needs to write what he or she wants to track and the app will either create a new log or adds a new record to existing one.

How we built it

Mainly prompting Bolt.new. Some visual debugging locally required but all the implementation was done by me prompting Bolt.new

Challenges

  • Make More Money Challenge
  • Custom Domain Challenge
  • Startup Challenge

Challenges we ran into

  • I learnt late that I can have Pro subscription for free when much work has been done;
  • Subsequent implementation depended heavily on the code that was written in the beginning;
  • Tokens burnt too quick so I had to sacrifice refinement of UI and code quality;
  • Since the begginning the project was written as only frontend application and I decided to keep it as it is because I was afraid of spending too much tokens. But I don't like the fact that secrets are in the js bundle;

Accomplishments that we're proud of

  • I directed Bolt.new to implement meta schema with supabase, and managed to make it implement tree rendering for records and forms and it works;
  • AI assistance works decently;
  • subscription works in test mode but it works!

What we learned

  • I should have been write rules right before prompting Bolt.new;
  • I should have been start from thinking on architecture but I relied too much on AI from the beginning;
  • How to spend tokens more effectively;
  • How to configure and work with Stripe and RevenueCat;
  • I should have been split the project into smaller submodules from the beginning;

What's next for Tracker Addict App

  • better UI, now it's too difficult in the form of trees, the forms are difficult to work with;
  • code quality, architechture improvement, refactor and optimization of requests;
  • move business-logic to backend;
  • make analytics reliable;
  • add AI assisted analytics;
  • add camera/photo/voice input and make it the shortest path for a user;
  • share logs schemas, logs presets;
  • add api for other apps to integrate;
  • make it platform allowing plugins for analytics and may be anything else...but for a while it's just very raw idea.

Built With

Share this project:

Updates