PreCog

Inspiration

What most inspired us to create our project was the fact that we live in a day of age where online shopping has become so popular and used by almost everyone at one point or another. After reading track 3, our idea became clear and we wanted to mimic online shopping tools like honey, but remove their controversial practices of paid promotions. Our program is similar in the fact that it exists to improve customer experience through recommending products based on customer patterns and preferability.

What it does

PreCog is a full-stack recommendation platform that connects a user’s Amazon account, syncs their purchase history, and turns that data into actionable shopping suggestions. The React frontend provides the sync dashboard, recommendation surfaces, and chat-style interactions, while the Go backend manages API routes, data validation, orchestration, and persistence in SQLite. During sync, the backend reads stored credentials from the Providers table, invokes a Python worker that fetches Amazon order history, and atomically replaces prior order records so failed syncs do not corrupt existing data. A dedicated search module queries Amazon listings through SerpApi and normalizes key product attributes like title, price, rating, image URL, and product link, which are then ready for the LLM layer to produce personalized, context-aware purchase recommendations for the user.

How we built it

They way the our project was built was split into two sections simultaneously working along side each other. Both a front-end and back-end was formed, as the projected progress, we bounced ideas to each other talking about problems and potential solutions to better incorporate each team's work.

Challenges we ran into

A big challenge we faced in the beginning was retrieving data to work with. After much time going over this problem, we overcame this issue by creating a custom environment that enabled us to ethically retrieve user data to use for the rest of the project.

Accomplishments that we're proud of

With the majority of our team having this being the first hackathon we attended, we are proud of not only the collaboration we achieved, but also the open mindsets and willingness to learn and overcome unfamiliar tasks such as working with React and making the frontend and backend work together for our final product.

What we learned

Similarly to as we mentioned earlier, this was 4/5 members first time attending a hackathon and because of this, we learned how to effectively be productive in groups. For example, we brainstormed as a group, solved problems as a group, and distributed work amongst each other to best work efficiency.

What's next for PreCog

Refine and build on existing features to more accurately find products to recommend and give users a more fulfilling experience when looking for specific products and allowing them to set budget restrictions, sort products by categories, etc.

Built With

• css
• go
• javascript
• python
• react
• serpapi