The Cellar

AI-Powered Stock Management for Pubs

Overview

The Cellar is a web application that helps pub staff speed up stock taking using AI-powered visual recognition. Point your phone at a fridge shelf, and the AI identifies products and counts quantities automatically.

Current Status

🚧 Phase: Planning & Prototyping

Currently in Phase 1 (PLAN) of the PSB system, defining requirements and validating the core AI scanning concept.

Problem We're Solving

• Manual stock taking in pubs is slow and error-prone (pen & paper)
• Staff spend too much time counting bottles and cans
• Inaccurate counts lead to ordering mistakes

Our Solution

• AI Vision Scanning: Point camera at shelf, AI identifies products and counts them
• Fast & Accurate: Complete stock counts in minutes, not hours
• Easy to Use: Works on any phone/tablet via web browser

Tech Stack

• Frontend: Next.js, Tailwind CSS, Shadcn UI
• AI/Vision: GLM-4V Vision API (Zhipu AI)
• Database: Supabase (PostgreSQL)
• Hosting: Vercel
• Auth: Supabase Auth

MVP Features

✅ Camera-based shelf scanning ✅ AI product identification ✅ Quantity counting ✅ Manual corrections ✅ Staff/Manager roles ✅ Basic stock tracking

Project Documents

• PROJECT-SPEC.md - Complete project specification
• How-I-Start-EVERY-Claude-Code-Project.md - PSB methodology reference

Development

Following the PSB (Plan, Setup, Build) methodology from Claude Code best practices.

Phase 1: PLAN ✅

• Define project goals
• Create product requirements
• Define technical stack
• Choose AI vision provider (GLM-4V)

Phase 2: SETUP 🚧

• GitHub repository setup
• Supabase project setup
• Environment variables
• CLAUDE.md configuration
• Project documentation structure

Phase 3: BUILD 📋

• MVP: AI scanning functionality
• User authentication
• Stock management dashboard
• Testing & iteration

License

TBD

Contact

For questions or feedback, please open an issue.