Makan Mana

Quick food discovery powered by AI - find spots fast, with real Google reviews

Comment

Inspiration

We've all been there - scrolling endlessly through Google Maps asking "makan mana hari ni?" We wanted to build something that lets you ask naturally in Manglish and get instant restaurant recommendations. Not replace Google Reviews, but make discovery faster.

What it does

MakanMana is an AI-powered food discovery app for Malaysian restaurants. Ask "nasi lemak sedap kat Bangsar" in Manglish, get instant recommendations with real Google reviews. Filter by budget and cuisine, save favorites, works on all devices.

How we built it

Frontend: React + TypeScript + Vite + Tailwind CSS

Backend: Supabase Edge Functions (Deno) with PostgreSQL

AI: Google Gemini 2.0 Flash for chat and image generation

APIs: SerpAPI for Google Maps search and reviews

Edge Functions: chat (main AI), vibecheck (review analysis), generate-food-image (AI images)

Design: Malaysian hawker center vibes with sambal reds and pandan greens.

Challenges we ran into

API rate limits, location logic complexity, Edge Function deployment, and database RLS setup. All solved with proper error handling and lots of iteration.

Accomplishments that we're proud of

✅ Complete AI food discovery app built in hackathon timeframe

✅ AI understands Manglish and Malaysian food culture

✅ Real Google Reviews integration via SerpAPI

✅ Beautiful Malaysian design

✅ Production-ready code

What we learned

Supabase Edge Functions, AI prompt engineering, SerpAPI integration, and the importance of proper error handling. Prompt engineering is harder than expected - small changes make huge differences.

What's next for Makan Mana

Deploy to production, mobile app, expand to other Southeast Asian countries, social features, and personalized recommendations. Vision: Become the go-to food discovery platform for Malaysia! 🇲🇾

Built With

Share this project:

Updates