Skip to main content

About the project

Tagline

Before you replace it, let’s understand it and see whether we can fix it.

What it is

LET’S FIX IT AI is an honest repair-diagnosis assistant for everyday products (car remotes, Bluetooth speakers, phones, and more). Users describe a problem (optional photo), get structured follow-ups, an evidence ladder (Observed / Possible / Unknown), real YouTube repair videos, and India parts search links (Amazon.in · Flipkart · Meesho) — without inventing prices, fake videos, or false confidence.

Prize tracks

  • Best Use of Gemma 4 — multimodal text + image diagnosis with open-weight gemma-4-26b-a4b-it via Gemini API; focused community repair tool
  • Best Open-Source AI Project — open-weight Gemma is core to how diagnosis works (structured JSON agent harness)

Live demo

https://lets-fix-it.vercel.app

How it works (minimal)

User chat (+ photo) → POST /api/diagnose (Hono) → Gemma 4 → Zod-validated JSON → UI (chat, Quick Answers, evidence, videos, parts)

Tech stack

  • Frontend: React 19, TypeScript, Vite, Tailwind (Material Design 3), React Router
  • API: Hono on Node (local + Vercel serverless)
  • AI: Gemma 4 primary (Flash Lite fallback only if Gemma fails)
  • Validation: Zod shared diagnosis schema
  • Integrations: YouTube Data API v3, merchant search URLs, optional Supabase auth/My Repairs
  • i18n: English + Telugu (UI + AI replies)

Demo path

  1. Try Car remote demo
  2. Answer Quick Answers
  3. Optional photo upload (multimodal Gemma)
  4. See evidence ladder → repair plan
  5. Open YouTube + shop links
  6. Switch Profile language to Telugu

Built for Hacktoberfest Hack Day Vizag × AI for Vizag.

Built with

reacttypescriptvitetailwind-cssreact.jshonogemmagemini-apizodfirebasesupabasevercelnode.jsmaterial-uijavascriptgemini

AI tools

Coded withCursor

Prize Categories Entered

  • Deepmind Google

    Best Use of Gemma 4

    Submitted