Garden

Social media but you control the algorithm

Comment

Inspiration

I'm inspired by the embarrassing amounts of time I spent on social media and the amount of times I've tried to quit, and have come up short. I'm in shock of how dystopian our world has become - we're all addicted to screens and spend literal hours of our day according to some algorithm rather than our intention, which has translated to a sense of general contempt, and inspiration for this project. I'm inspired by apps like Notion, The Browser Company and other tech companies who have took status quo applications (notetaking apps, the browser) and made something magical out of them in a way that makes it hard to look back on the old way to do things, and that inspired me to do the same for social media.

What it does

Garden lets you prompt your social media feed so you spend your screentime according to your intention, not an algorithm. If you're at the gym, you can tell your phone that, and it'll send content that'll keep you focused. If you're studying, you can tell your phone to avoid brainrot. You can scroll on all your favourite content types (Instagram, Reddit, Tiktok, X, YouTube) on Garden through official APIs and Pinterest like content aggregation, in an environment dictated by your intention rather than manipulative algorithms.

How we built it

SwiftUI, Supabase as a backend, OpenAI for vector emebddings, and a lot of caffeine

Challenges we ran into

Content scraping and data cleanup was tough, half of our data ended up being dummy data. Furthermore, significant thought was given to the architecture, as creating a centralized social media feed required working across various modes. Furthermore, to vector index the content, it was challenging finding an effective multimodal embedding tool. Lastly, I spent a large amount of time on storytelling for the pitch deck, while balancing this grand vision with objective data.

Accomplishments that we're proud of

Effectively using the new Liquid Glass API was really fun, and added a modern stylish flair to the app. Also, generally proud of myself for getting both a technical MVP and a pitch deck done this week as a solo hacker :)

What we learned

Content scraping from various sources is challenging! Getting them onto one format, and cleanly accessing the data required me to test several solutions. On the contrary, vector embeddings were relatively easier than I thought they would be, though this may have to do with the relative scope of the project - creating a full semantic search system is likely a cutting-edge technology problem.

What's next for Garden

My intention with Garden is to take what I incubated during SpurHacks and turn it into a full-fledged tech startup. Beyond this weekend, I will apply to the SpurHacks accelerator, have begun the application process for Front Row Ventures, and will actively recruit my first users for the closed beta, as outlined in phase 1 of the project. I carry a strong belief that the problem Garden addresses is one of the defining problems of our generation, and am incredibly motivated to make a contribution towards a solution for it.

Built With

Share this project:

Updates