Rooted

Please see this video instead https://www.loom.com/share/c650ba5c28104bbabdda1c0fb117ed31

Inspiration

As someone from Sacramento, I kept seeing the city's slogan, "America's Farm-to-Fork Capital," everywhere, especially on the water tower I passed all the time. But the more I thought about it, the more I questioned what that slogan actually meant in practice. If a product says it is organic, local, or farm fresh, how does a regular shopper know that is really true? How far did it travel? Was it actually grown nearby, or just marketed that way?

That question became the starting point for Rooted: "What if there were a way to check whether your food is really farm to fork?"

That idea led me to build Rooted, a product focused on making food transparency simple, visible, and useful. Instead of asking people to blindly trust packaging and branding, Rooted helps them see the story behind their food in a way that feels immediate and understandable.

What it does

Rooted is a food transparency and verification tool that helps users understand where their food came from and whether its story holds up. It:

Scans a product or batch and turns it into a readable results page
Shows the farm source, distance traveled, soil quality, and a calculated Trace Score
Uses the Gemini API to generate a short, plain-English summary of the product's background
Surfaces nearby local alternatives when the current product has a weaker route or footprint
Uses Browser Use agents to check whether a source, such as a farm, has an up-to-date certification in the government's Organic Integrity Database
Connects transparency with action by exploring ways to support growers more directly

Rooted is meant to close the gap between branding and reality. Instead of just hearing "local" or "organic," users get a clearer picture of what those claims actually mean.

How we built it

Frontend

React for the app UI
JavaScript for component logic and app behavior
styled-components for component-scoped CSS inside the React code
HTML through JSX/React components to render the results page UI

Backend

Node.js server for routing data and coordinating product lookups
Supabase for caching product, market, and verification-related data
Distance calculation and geolocation logic for the miles-to-source feature

Verification + AI

Gemini API for generating the short product summary
Browser Use for checking certification-related information and finding nearby local alternatives by reading farmers market websites
Solana for a gratitude system that sends support directly back to farmers

Challenges we ran into

Trustworthy AI output: Sounding overly positive or making up information goes against the goal of transparency. We had to simplify the prompts and add restrictions to make sure users were getting clear and accurate information.
Verification latency: Live checks are useful, but they can slow the experience down if they are not handled carefully. We optimized calls to the Browser Use agent to reduce verification time and lower credit usage. We also added a fallback using known information from previous scans in the Supabase database for reliability and better performance.
Messaging balance: We wanted Rooted to feel honest and informative without becoming overly technical or alarmist. We kept the information concise while still showing what the system is doing. For example, when finding alternatives, the Browser Use agent's actions are displayed on screen.

Accomplishments that we're proud of

Built a working end-to-end experience that turns a scan into a meaningful food transparency result
Combined AI summaries, verification, route distance, and a custom Trace Score into one product flow
Created a system that can challenge weak food claims instead of just repeating them
Added local alternative sourcing to make the product actionable, not just informative
Implemented a functional Solana donation interaction that lets support add up for the farmer bit by bit