AI / Vendor pricing
Sheet Parser
An AI application that turns vendor price sheets into structured pricing data, keeps a shared vendor catalog, and organizes prices by project.
- React.js
- Node.js
- TypeScript
- OpenRouter
- LLMs
- Prisma
- PostgreSQL
Project scope
Connects React and Node.js with OpenRouter LLMs for price-sheet parsing, with vendor and project data stored through Prisma in PostgreSQL.
The problem
Vendors organize price sheets differently. Teams need to turn those sheets into usable records, retain the vendor context, and find the prices relevant to each project.
- Vendor price sheet
- LLM parsing via OpenRouter
- Shared vendor catalog
- Project pricing
Illustrative workflow based on the project description; no private price sheets or customer data are shown.
How it works
The application uses React.js and TypeScript for the interface, Node.js to connect with LLMs through OpenRouter, and Prisma with PostgreSQL to store pricing and its vendor and project relationships. The workflow links price-sheet interpretation to a shared catalog and project pricing.
LLM price extraction
Use models through OpenRouter to interpret price sheets and extract pricing data.
Vendor records
Keep parsed prices linked to the vendor that supplied them.
Shared vendor catalog
Maintain vendor pricing in one catalog across the workspace.
Pricing by project
Associate relevant vendor prices with individual projects.
Engineering considerations
LLMs can help interpret varied sheet layouts, but extracted prices still need to be checked against the source. Missing values, currency, and ambiguous rows require validation. Vendor and project relationships also need to remain explicit in the stored data.
What this enables
Teams can organize pricing by project while retaining a shared view of vendor data across the workspace.
Related engineering notes
Vivek Jalondhara
Full Stack Software Engineer