Side Project · Building Now

butter
Menu

better menu — but made by butter (that's me 🧈). A food discovery platform that started with one real problem: my mom's restaurant has no website.

The Pitch

"Every time I wanted to eat out, I had to guess. No menu online. No idea what's good. At some point I stopped waiting for someone else to fix it. First restaurant? The one closest to home— my mom's!"

🍜 Starts with mum's restaurant 🤖 AI from menu data — no users needed 🛒 Ordering to help mum's knee pain
🍜 ครัวแม่
Wiang Haeng · Open now
TH/EN
🤖
ไม่รู้จะกินอะไร?
ให้ AI แนะนำจากเมนูนี้
ลอง →
🍳
ข้าวผัดกะเพรา
Basil Fried Rice · 🌶️🌶️
฿65
🍲
ต้มยำกุ้ง
Spicy Shrimp Soup · 🌶️🌶️🌶️
฿120
🥢
ผัดไทย
Pad Thai · 🌶️
฿80
🥗
ส้มตำ
Papaya Salad · 🌶️🌶️🌶️
฿60
Scan QR · No app needed ✿
📋 Case Study

Why I built it this way

Every decision here has a reason. This is what separates building a product from building a feature.

😤

The Problem I Had

I wanted to check a restaurant's menu before going. But most places don't have a website. I'd scroll Google Maps reviews hoping someone photographed the dish I wanted. It was annoying every single time.

😓

The Problem My Mum Had

She runs the restaurant alone. Foreign tourists can't read Thai menus. She loses customers who give up. She has knee pain and walks to every table to take orders. Two problems, both real, both fixable.

💡

The Real Success Metric

Not DAU. Not revenue. Not conversion rate.

Mum walks less. Mum's knees hurt less.

If that happens — the product works. Everything else is a bonus.

🗺️

Why Start With One Restaurant

Cold start kills two-sided platforms. The fix: prove it works for one real user first — then walk into any other restaurant with a live demo. Mum's restaurant is the pitch, not just the prototype.

Product Decisions — and why
✓ Build
Ordering via QR before AI recommendations — Because mum has knee pain. Reducing how much she walks is immediate, measurable value. AI is cool but doesn't fix her knees.
✗ Skip
Real-time payment processing — Adds friction and complexity for zero extra value right now. Mum's customers pay cash at the table. This can come later.
✓ Build
AI recommend from each restaurant's menu only — Works with zero user history. Day one, a customer scans QR → gets 3 relevant dish suggestions. No cold start problem for the AI feature.
✗ Skip
Cross-restaurant discovery (Phase 1) — Meaningless without restaurant data. Build supply first. Turn on demand-side features when there's something to discover.
✓ Build
Pre-order for group bookings — Mum's restaurant does catering for groups. They book in advance. Pre-order means she knows exactly what to prepare, less waste, less stress.
✗ Skip
Multi-tenant self-serve signup (Phase 1) — Phase 1 is one restaurant. Mum. I setup everything personally. Self-serve comes when I have a live demo to show other restaurants.
📋 Build Log

What's been built. What's next.

Updated as features ship. A living record, not a pitch.

Actively building
Shipped
🌱

Nothing shipped yet —
but that's about to change.

🔨 In Progress
QR menu — bilingual TH/EN
building now
Admin dashboard — CRUD menu + image upload
building now
AI translation — Thai name → EN description
building now
🗺️ Roadmap

Supply first. Then demand.

Cold start kills two-sided platforms. The fix: start with one restaurant, prove it, then scale with a live demo in hand.

🍜
Phase 1 · Now
One Restaurant. Prove It Works.
Mum's restaurant in Wiang Haeng. QR menu + AI recommend + ordering. Real user, real problem, real feedback loop. No pressure to scale yet.
🔨 Building Now
🏪
Phase 2 · After validation
Open to Any Restaurant.
Walk into restaurants in Pai and Chiang Mai with a live demo — mum's page. Self-serve signup. Every restaurant gets their own subdomain and AI recommender.
📋 Planned
Phase 3 · When restaurants exist
Food Discovery for Users.
"What should I eat today?" — now we have real menus from real restaurants. AI recommends across the platform. Browse menus before you go. User accounts + taste profiles.
💡 Vision
🛒
Phase 4 · Future
Order from the Table. Order Before You Arrive.
Full ordering flow for restaurants that need it. Pre-order as a member. PromptPay + card. Kitchen display system. By this phase — users trust the platform.
💡 Vision
⭑ Tech Stack

Built to scale from day one

Next.js
Frontend
Expo
Mobile
🐹
Golang
Backend Phase 3+
🤖
Claude AI
Translation + Recs
🐘
PostgreSQL
Database
WebSocket
Real-time orders
🐳
Docker
Infra
💳
PromptPay
Payments Phase 4