We delivered an AI estimator for a custom cabinetry shop on Johns Island — one that prices jobs from architectural drawings by calculating material, labor, and finishing costs against current supplier pricing and shop capacity, tracks lead times for specialty hardware and sheet goods, and generates proposals detailed enough to win work from interior designers without leaving margin on the table

He started building cabinets in his garage twenty-two years ago, and now he runs a six-man shop on Johns Island that does custom cabinetry for high-end residential projects in Charleston, Kiawah, and Bluffton. His work ends up in houses that cost between two and eight million dollars, designed by architects and interior designers who specify things like rift-sawn white oak with a cerused finish, integrated LED channel lighting, waterfall edge countertop returns, and soft-close European hinges that cost fourteen dollars apiece. He builds beautiful work. He does not price it well. Every estimate started the same way: an interior designer or general contractor would send a set of architectural drawings and a finish schedule, and he would sit at his desk for two to four hours with a calculator, a sheet goods price list that was three months old, and a yellow legal pad. He would count doors, drawers, and linear feet of countertop. He would estimate sheet goods by mentally nesting the cabinet boxes onto four-by-eight sheets of plywood, which he had been doing long enough to get within ten percent, but ten percent on a kitchen that uses forty sheets of walnut plywood at two hundred and eighty dollars a sheet is over a thousand dollars of margin that could go either direction. He would guess at labor hours based on how the last similar job went, but the last similar job was six months ago and he had a different crew then. The finish cost was the hardest part because a hand-rubbed lacquer finish takes three times the labor of a catalyzed conversion varnish, and he could never remember exactly how many hours per door the lacquer had taken on the Sullivan's Island kitchen versus the Daniel Island one. He would add it up, add fifteen percent, look at the number, decide it was too high, take off five percent, and send the proposal. He won about half the jobs he bid. He lost money on about a third of the ones he won. We built him an estimator that prices the way he thinks — in cabinets, not in spreadsheet cells. When a set of drawings comes in, the system breaks the job into individual cabinet units: base cabinets, wall cabinets, tall cabinets, islands, built-in pieces, and specialty items like range hoods and appliance garages. Each unit has a material takeoff: so many square feet of face frame stock, so many square feet of panel material, so many doors of a specific style and size, so many drawer boxes, so many linear feet of crown or light rail molding. The sheet goods calculation is precise — the system nests parts onto standard sheet sizes and accounts for grain direction, which matters for rift-sawn and quarter-sawn material because you cannot rotate parts freely the way you can with plain-sawn. Material costs update monthly from his three primary suppliers. He buys domestic hardwood plywood from a distributor in Savannah, European hardware from a specialty importer in Atlanta, and solid lumber from a mill in McClellanville. When the Savannah distributor raised walnut plywood by twelve percent in March, every estimate the system generated from that point forward reflected the new price. He did not have to remember to update a price list. When the hardware importer's lead time on Blum Aventos lift systems went from three weeks to nine weeks because of a shipping delay, the system flagged every active proposal that included lift systems and adjusted the project timeline. Labor estimation works from his actual history. Every job his shop completes gets a final hour count by task category: box construction, face frame assembly, door and drawer front fabrication, finishing, hardware installation, delivery, and site installation. The system knows that his crew builds frameless Euro-style boxes faster than face-frame traditional boxes, that his finisher takes 1.4 hours per raised-panel door for a hand-rubbed lacquer but only 0.3 hours for a sprayed conversion varnish, and that site installation in a downtown Charleston house with no elevator and a third-floor kitchen takes twice the labor of a ground-floor installation in a new-construction house on Kiawah with a twelve-foot-wide front door and no stairs. The proposals themselves solved a problem he did not know he had. His old proposals were a number on a page — "Kitchen cabinetry per plans: $47,000." Interior designers hated this because they could not see what they were paying for, and when the client asked why the cabinets cost forty-seven thousand dollars, the designer had no answer except "that's what the cabinet guy said." The system generates itemized proposals that show each cabinet unit, its dimensions, its material, its finish, and its cost — not broken down to the penny where it would reveal his margins, but detailed enough that the designer can see that the ten-foot island with the waterfall walnut end panel costs more than the two standard base cabinets flanking the sink, and can make informed decisions about where to spend and where to save. His close rate went from roughly half to over seventy percent in the first four months. His average margin improved by eight points because he stopped under-pricing the complex jobs — the ones with curved doors, book-matched panels, and hand-applied finishes that he used to estimate by gut and always got wrong in the same direction. "I used to spend Monday mornings doing estimates," he said. "Now I spend Monday mornings in the shop, which is where I should have been the whole time. The estimates are better than the ones I did by hand, and I have not left money on a proposal since February."