AI Menu Integrator
A multi-tenant service that takes a restaurant's menu as PDFs or photos and turns it into structured menu data a point-of-sale system can import.
- Status
- In development
- Role
- Developer
- Year
- 2026
- Type
- Menu digitisation SaaS for restaurants and POS systems
Where it stands
Accounts, organisations, roles, restaurants and menu import are built: uploaded files are rendered page by page and their text is extracted, with OCR for scans. Turning that text into products and prices, the menu editor and the exports are not built yet.
There is no public link for this project.
My part
Architecture and build.
- Built accounts, organisations, members and roles, and restaurants with locations
- Built upload and menu import for PDF, PNG, JPEG and WebP files
- Built the background worker that renders pages and extracts text, using the file's own text layer where there is one and OCR for scans
- Isolated every organisation's data with row-level security
What it is
Restaurants keep their menus as PDFs and photos. Point-of-sale and delivery systems need them as structured data: items, prices and options. Retyping is slow and goes wrong.
The service is a modular monolith: a Next.js web app, a FastAPI service, PostgreSQL as both the database and the task queue, and separate worker processes for document handling. Every record is isolated per organisation with row-level security.
Today it gets as far as every page of every uploaded file rendered to an image, with its text extracted and positioned. Progress, per-page results, failures, retry and cancel are all visible in the app.
Built with
Python · FastAPI · Next.js · PostgreSQL · Row Level Security · Alembic · OCR
Need something like this built?
Tell me what you are making. I will tell you how I would approach it, and whether I am the right person.
Start a conversation