An AI-based ERP that runs garment exports from tech pack to order closing
TB-ERP reads buyer tech packs and POs with AI, then carries every order through costing, TNA, sourcing and order closing in one system.
- Time saved on costing, bulk, tracking and management
- 80%
- Accuracy from AI cross-checking
- 99%
- Connected ERP modules
- 25+
- Client
- TB-ERP
- Industry
- RMG
- Timeline
- 2025 to present
- Platforms
- Web app, AI service, background jobs
- Services
- Product design, Web platform, AI solutions, Cloud and DevOps
The challenge
Garment merchandising ran on spreadsheets. Every buyer tech pack and purchase order was retyped by hand into costing sheets, booking files and TNA calendars. Fabric consumption came from CAD in one file and labour costs from IE in another. Every copy was a delay and a chance to make an error, and mismatches between the PO and the tech pack often surfaced only at sampling or shipment.
What we built
- AI tech pack and PO reader. Reads PDFs, scans and photos as one document and extracts styles, BOM lines, colourways and size quantities into a strict schema, skipping anything cancelled or crossed out.
- AI cross-checking. Compares the PO with the tech pack style by style and size by size, and flags differences before costing starts.
- Costing. CAD fabric consumption and IE costing imports, bulk and average costing sheets, and shared fabric and accessory libraries with supplier rates.
- Planning and sourcing. TNA calendars, a daily tracker, sample requisitions, nominated suppliers and supplier purchase orders.
- Commercial and closing. Sales contracts, budgets, approvals and order closing reports.
How we worked
The ERP is split into a .NET core, a React web app and a separate Python AI service. The core follows Clean Architecture with automated architecture and integration tests, and long-running work such as imports and reminders runs as background jobs, so the web app stays fast.
Our merchandisers used to retype every tech pack and PO into Excel before costing could start. Now TB-ERP reads them, cross-checks the quantities, and costing, TNA and supplier orders all run from the same data.
Technology
- React
- TypeScript
- Tailwind CSS
- .NET
- SQL Server
- RabbitMQ
- Hangfire
- Python
- FastAPI
- Gemini
- Docker
Planning something similar?
Tell us what you want to build, improve or automate. We reply within one working day.