Car dealerships
Quintessence Auto: a dealership back office that updates itself
Client
Quintessence Auto Group
Sector
Automotive dealership
Location
Montréal, Québec
Built
April 2024, three to four weeks
The client
Quintessence Auto Group is an automotive dealership operating out of Montréal, Québec. Like most independent dealerships, their inventory is the business: vehicles come in, get photographed, described, priced and listed; details change as units are inspected or reconditioned; and sold cars need to come down before a buyer calls about something that left the lot last week.
On the acquisition side, their sales reps work auction channels, scanning listings, comparing units, deciding fast what is worth bidding on. Both sides of that operation are information work: the same vehicle data keyed and re-keyed into different places, the same auction pages read line by line.
In April 2024, Quintessence brought us in to rebuild that layer, the website their customers see and the back office their staff live in, as one system instead of a collection of manual chores.
The problem
Dealership inventory work is repetitive by design. Every new unit means photos, specs, trim details, mileage, pricing and a listing written up and published; every change, a price drop, a sale, a correction, means finding the listing and editing it again. When the website is a separate thing from the system staff actually use, that work doubles, and the site quietly drifts out of date.
Auction sourcing has the same shape: reps spend hours reading through raw auction listings in formats that were never designed for quick comparison, extracting the few fields that matter to a buying decision. None of this is skilled work, but all of it eats the hours of the people whose actual job is selling cars.
This build set out to remove that manual layer: listings that publish and update themselves from one place, and auction data arriving already parsed instead of needing to be read.
- Every unit keyed in twice. Vehicle details, photos, specs and price were entered for the back office and again for the website, then corrected in both every time something changed.
- Listings drifted out of date. With the site separate from the system staff actually used, sold units lingered and prices went stale.
- Auction sourcing done by eye. Reps read raw auction listings line by line to extract the few fields that drive a bidding decision.
- Skilled people on unskilled work. None of it needed a salesperson, and all of it consumed the hours of people whose job is selling cars.
Constraints
The build ran over three to four weeks in April 2024. Both halves had to ship together, a customer-facing website is only useful to the dealership if the backend feeding it removes work rather than adding a second place to type.
The auction dashboard had to work with auction sources as they exist, which meant AI scraping and parsing of listing data that arrives in inconsistent formats, normalized into something reps can scan in seconds. Everything had to be usable by sales staff without training or a manual.
What we built
Custom dealership website
A fast, custom-built site presenting the dealership's inventory to buyers. Listings are rendered directly from the backend's data, so what a customer sees is always what the dealership actually has, no separate publishing step, no stale pages for units that already sold.
Inventory backend dashboard
The single place staff manage vehicles. Adding a listing, updating details, changing a price or marking a unit sold happens once, here, and flows through to the website automatically. The dashboard replaces the re-keying and copy-paste that previously connected the back office to the site.
Automated listing pipeline
The machinery between the dashboard and the site: when a vehicle record is created or edited, the public listing is generated and updated without anyone touching the website itself. Time-consuming listing admin, the part of the job nobody was hired for, became a side effect of normal record-keeping.
AI auction scraping and parsing
An AI layer that scrapes auction listings and parses the unstructured data, vehicle details, condition notes, the fields a buyer actually weighs, into clean, structured records. It does the reading so reps do not have to.
Auction dashboard for sales reps
A custom dashboard built on top of the parsed auction data, giving Quintessence's sales reps a scannable view of what is on the block. Instead of working through raw auction pages one at a time, reps evaluate normalized listings side by side and spend their attention on the buying decision.
