She spent nine years buying produce for a restaurant group downtown, which is how she learned that the microgreens on a twenty-eight-dollar plate had usually been on a truck out of Florida for three days. In 2022 she leased a half-acre and a metal building off Bears Bluff Road on Wadmalaw Island and started growing them herself. The operation is smaller than people expect. One insulated grow room, twenty-four by forty-eight feet, holding six racks of five tiers, eight trays to a tier.
Two hundred and forty tray positions, and every one of them is either earning or it is not. She grows nineteen varieties on cycles that run from six days for radish to sixteen for amaranth. Sunflower takes nine. Pea takes eight.
A tray of sunflower wants three hundred grams of seed; a tray of amaranth wants twelve. Miss the seeding date by two days and the crop comes out fibrous and bitter, or too short to cut at all. Her customers are thirty-four restaurants between King Street and Shem Creek, plus a Saturday stall at the Mount Pleasant Farmers Market. Chefs commit to standing orders and then change them.
A sous chef texts on Monday asking to double the pea shoots for a Thursday tasting menu, but pea needs eight days in the rack, so Monday is four days too late. She could either say no, which chefs remember, or she could carry enough surplus that she almost never had to say no, which is what she did. That surplus was the whole problem. She planned production on Sunday nights with a legal pad and a spreadsheet she had built in her first year, working backward from a standing-order sheet that was accurate for about the first two days of any week.
To protect herself she over-seeded roughly a fifth of the room. In her last twelve months of doing it that way she cut and composted eighteen percent of what she grew, which is seed, water, power, coir, and her own labor going into a wheelbarrow. Seed alone ran about twenty-nine hundred dollars a month. She was also spending five or six hours every Sunday building a plan that a single Tuesday phone call could unmake.
What we built for her is a production planner that treats those two hundred and forty tray positions as the constrained resource they actually are. It holds fourteen months of her invoices, so it knows that the Broad Street account orders radish every week and amaranth only when there is a private dinner, and that market sales fall by a third when Saturday rain is forecast before Wednesday. It back-schedules every seeding date from committed orders plus a demand forecast per variety, checks each proposed date against the crop time for that variety, and adjusts for the grow room's actual germination behavior — she runs two degrees warm in August, which pulls radish in half a day early and matters when a chef wants it Friday and not Thursday.
Every morning it prints a seeding list: variety, tray count, seed weight in grams, rack and tier assignment, and the order each tray is already spoken for. It prints a harvest list the same way. When she weighs a finished tray, she enters the yield, and the planner reconciles actual against planned and carries the correction into the next cycle, so its estimate of what a tray of sunflower gives her in September is built from her room in September and not from a seed catalog. When a chef calls in a change, she can see in a few seconds whether there is an open position and enough runway to say yes, and if there is not, it tells her what she could offer instead that is already in the rack and cutting that day.
She has been running it for four months. Waste is down from eighteen percent to a little over four. Seed spend has fallen to about twenty-one hundred dollars a month. She has taken on nine new accounts without adding a single rack, because the positions were always there and she had been using a fifth of them as insurance against not knowing.
The Sunday night planning session is gone. "I thought I needed a bigger room," she said. "I needed to know what was already in the one I had."