Instant quoting and order management platform for made-to-order products

A made-to-order product can't have a price list: price depends on configuration, materials, finish, measured dimensions and destination, and freight on heavy palletised goods is a large, volatile share of the total. Quoting by hand meant a salesperson, a phone call to a carrier and a day or two's wait, while buyers pricing several suppliers at once went with whoever answered first. None of that freight math is possible before the goods exist — a carrier won't quote without a freight class, a pallet count and a stacked height, none of which exist until something has been configured. Once an order was placed, the work didn't stop: purchase orders, supplier invoices that rarely match them exactly, shipment tracking buried in email attachments, and a growing back office to handle it all.

Built freight as a first-class domain calculation — deriving weight, freight class, pallet count and stacked height from the chosen configuration and materials — then rated several national carriers live in parallel, each carrier implemented as its own class behind a shared interface so the pricing engine compares options without knowing how any individual rate was obtained. Landed cost is assembled from component pricing, finish, fees and the selected freight option, with margin applied through a single governed rule with a protected floor rather than left to individual judgement.

Treated inbound email as a system input rather than a mailbox: supplier invoices, carrier tracking notifications and customer replies are ingested, threaded to the right order, and parsed — invoices text-extracted from PDF and matched against the originating purchase order with the variance stored as data, and out-of-tolerance cases routed into a typed exception workflow instead of an inbox. Tracking numbers and carrier identity are recovered by pattern extraction from PDF attachments on carrier email and used to trigger customer notifications automatically. All of it runs behind an explicit order state machine with per-stage task queues, and scheduled work — follow-ups, decline processing, daily reporting — is guarded by a database-level semaphore so the platform scales across instances without duplicating a single email.

Buyers get a freight-inclusive price in minutes instead of days, priced from real carrier rates rather than estimates, with pricing consistent regardless of who or what generated the quote. The back office scales with automation rather than headcount: purchase orders, invoice matching, tracking extraction and accounting sync all happen without manual re-keying, discrepancies surface as workflow instead of being found late, and every order has a visible owner and stage.

Something holding you back?

New build, rebuild, or adding AI to what you already have — I'll take a look and give you an honest perspective.