QuoteDesk
Quoting automation for bespoke interior options on an ultra-premium automotive programme.
- Role
- Design, build and deployment
- Period
- 2025 – 2026
- Status
- Live
- Stack
- Next.js 14, TypeScript, Supabase, PDF generation, Vercel
The problem
Each vehicle on the programme arrives with a customer specification pack listing hundreds of parts and options. Quoting it by hand meant cross-referencing several versioned spreadsheets, applying standards that differ part by part, and re-keying everything into a document. It took hours per car and errors were only caught downstream.
What I built
- Imports the specification pack directly and maps each line to a known part, falling back to description matching when the OEM renumbers parts, and learning new numbers as it goes.
- Rules-based pricing engine that encodes which finishes and options are standard, included or chargeable for each part, with a certainty score on every auto-proposal so a reviewer sees what needs a second look.
- Versioned cost matrices so a re-quote against a newer revision is a diff, not a rebuild.
- Generates a branded PDF quotation and keeps a full audit trail per job.
Outcome
A car that took 4 hours to quote now takes 30 minutes, with review effort concentrated on the handful of lines the engine flags. 30 to 40 people across commercial and engineering use it. Pricing rules live in one place instead of in people's heads.
Customer, part and pricing data are confidential and not shown here.

