How an AI Quote Desk Priced Rentals From the Rate Sheet

The quote that depends on who wrote it

The company rents and builds equipment for industrial job sites: portable pads in set sizes and wall heights, plus fabricated units built from a bill of materials. Every rental quote started the same way. A rep opened the rate spreadsheet, found the row, worked out the term, and typed line items into the CRM deal.

When I checked real closed jobs against the company's own list price, the answers were all over the place. Two went out 23% under list. One twelve-month job went out 24% over list. Same spreadsheet, same kind of equipment, no consistent answer. Nobody was doing anything wrong on purpose. There was just nothing between the rep and the number.

What the rep does now

The rep types the request on the deal the way they would say it out loud: "40x40 wash pad with 8 foot walls for 2 months." They set the deal to request a quote and keep working.

About a minute later the deal updates. It shows the status (processing, then ready for review), the quote amount, the rental line item with the term built in, and separate lines for installation, removal and freight that the rep fills in. A manager then approves the quote or marks it as needing a correction, and that correction is saved for the next run.

Where AI helps and where it is kept out

AI is good at reading messy requests. Reps write typos, spell out numbers, use slashes for dimensions, mix quote types, or say "same as last time." The AI turns all of that into a few plain fields: size, wall height, term and quote type. A backup model takes over if the first one fails.

AI is not allowed near the math. Every rental price is looked up from a 40-row approved rate library, the same list the company already used. Fabrication quotes come from the bill of materials and a set margin. Using the company's own assembly costs, I reconstructed the labor hours for 11 of 13 fabricated units to within a few dollars of their real unit cost.

Two guardrails came out of live testing:

• Closest approved size. If a rep asks for a pad that is not on the sheet, the quote desk picks the closest approved size that covers it and says so on the quote.
• Stop on missing details. If the wall height is missing, the deal is marked as needing information instead of being priced on a guess.

What was tested

• 15 of 15 messy-request cases passed: typos, spelled-out numbers, slash dimensions, vague asks, mixed quote types, wall heights the company does not offer, and "same as last time."
• 5 of 5 real job specs, copied word for word from past deals, came back with the right item, term and line items.
• Every recent test deal the client ran in their CRM priced to a real approved rate.

What was not measured

I have not measured time saved per quote, how many reps use it day to day, or what happened to margins after launch. The pricing spread above is from closed jobs before the quote desk existed. Whether that spread narrows depends on how the team uses the review step, and I have no data on that yet.

Who this fits

This fits dealers, rental yards and fabricators where reps quote from a spreadsheet and the CRM only sees the final number. If you pulled your last ten closed quotes and compared them to your own list, and the gap surprised you, that is the problem this solves.

See how this works for equipment dealers, or start with a workflow scan to see where your quotes drift.

What I would revisit

I would show each rep the gap between the list price and the price they are about to send, right on the deal. Most discounts are not decisions. They are habits that nobody sees until the year-end numbers come in.