Built for an industrial equipment dealer in South India. It replaced hand-typed Word quotations with an instant, accurately priced PDF, and it's now how the business sends every quote.
Customers pick from 379 lifting products, and the system prices them, builds a branded quotation and emails both sides. I scoped it, turned a messy supplier price list into clean data, built it and shipped it.
Customers asking for a hoist or a pulley block had to wait while someone built the quotation in Word. Each one took about 20 minutes.
The prices lived in a 56-page supplier price list PDF, with a different table layout for almost every product family. Finding the right model meant matching capacity, lift height, suspension type and series by eye, then copying the price, the tax line, the standard terms and the next reference number into a letter.
It was slow, easy to get wrong, and it tied up the people who should have been selling. In a market where the first accurate quote often wins the order, every hour of delay mattered.
Speed wins orders. A customer who gets an accurate quote in minutes is more likely to buy.
The goal wasn't a prettier letter. It was to remove the waiting. If the system could find the right product and price it correctly on its own, the team would only need to follow up.
QUOTATION
SUB: SUPPLY OF CHAIN PULLEY BLOCK
Prices: GST @18% extra. Validity: 30 days.
Quotation emailed to customer and sales team
Illustrative example. Prices and reference numbers are not real.
The hardest part wasn't the form. It was getting clean data to power it.
Dozens of tables, each with its own column headers, units and footnotes.
I used an AI extraction agent to map every table to one schema, producing 379 products with a code, capacity, lift height, suspension type and price.
Cascading filters narrow the list as you go, so nobody has to know a model number.
The server looks up every price by product code and totals the quote. The browser never sends a price, so nothing can be edited on the way.
The quotation is laid out on the company letterhead with the tax line, standard terms, validity and a sequential reference number.
The customer gets their quotation. The sales team gets a copy whose replies go straight to the customer, so the conversation continues in one thread.
Built with Next.js, TypeScript, React Hook Form and Zod, PDFKit, Nodemailer and Firebase, deployed on Vercel.
The generated PDF reproduces the existing letter: the same letterhead, tax line, delivery terms and 30-day validity. Customers saw nothing unfamiliar, and the team didn't need retraining.
ResultEvery quote now goes through the system.
A form is only as good as the prices behind it. I spent the most effort normalising the supplier PDF into one clean schema before building any screens.
Trade-offSlower start, but no wrong prices in front of customers.
The plan was a live database for prices. Prices only change when the supplier issues a new list, so I bundled them with the site instead. It's faster, has no usage limits and can't go down.
Trade-offA price update needs a redeploy, which happens a few times a year.
Customers only send product codes and quantities. Every price is looked up and totalled on the server, so a quote can't be altered before it's generated.
Trade-offNone worth mentioning. It's the only safe option.
The business had no custom email domain set up, so I sent quotations through its existing mailbox rather than adding a new email service. Replies land where the team already works.
Trade-offA daily sending limit, comfortably above their volume.
Reference numbers run in sequence from a database counter, matching the business's existing numbering. If the database is ever unavailable, the system falls back to a temporary number so the customer still gets a quote.
Trade-offA rare out-of-sequence number, in exchange for never losing a quote.
More than 1,500 quotations have gone through the system since launch, at about 2 minutes each instead of 20. That's roughly 450 hours of sales time handed back to the team.
The number I'm proudest of is adoption. Internal tools often get used for a few weeks and then quietly abandoned. This one replaced the old process entirely, because it kept everything the team already relied on and removed only the waiting.
Keeping the familiar letter, terms and numbering is why the team switched completely instead of running both processes side by side.
The form took days. Getting a trustworthy price list out of a 56-page PDF was the real product problem.
"Minutes per quote" meant more to the business than any technical metric, and it made the value obvious on day one.
Bundled prices and the existing mailbox let it go live in weeks, without waiting on new accounts or infrastructure.
Built for a real business and used every day. The company name, supplier brand and all customer data are left out on purpose.