We delivered an AI shrimp dock operations manager for a third-generation Shem Creek seafood wholesaler who coordinates twelve trawler captains, three cold-storage facilities, and forty-plus restaurant accounts across the Lowcountry — one that forecasts catch volumes from tide tables and water temperature data, automates dock-to-door chain-of-custody tracking, manages dynamic restaurant pricing based on daily supply, and produces real-time inventory dashboards so nothing sits in a cooler longer than it should

His grandfather started the business in 1971 with a single boat and a handshake deal to sell white shrimp to three restaurants on East Bay Street. His father expanded it to a fleet of six boats and a proper cold-storage building on Shem Creek. He took over nine years ago and now coordinates twelve independent trawler captains — none of whom are employees, all of whom are stubborn — three refrigerated warehouses totaling fourteen thousand square feet of cold-storage capacity, and a customer list of forty-three restaurants, seven grocery stores, and a growing direct-to-consumer box program that ships Lowcountry shrimp to doorsteps from Greenville to Atlanta. The business runs on a margin so thin that a single bad decision about inventory or pricing can turn a profitable week into a loss. And he was making those decisions from the cab of his truck at four-thirty in the morning, squinting at text messages from captains who spell about as well as they type, trying to figure out how much shrimp was coming in today, where to put it, who needed it, and what to charge. The fundamental problem with fresh shrimp is that it is perishable in a way that most people do not appreciate. From the moment the net comes out of the water, the clock starts. Shrimp that sits in a cooler for forty-eight hours is still safe to eat but it is not the same product as shrimp that was swimming six hours ago. His best restaurant accounts — the ones that put "Shem Creek shrimp" on the menu and charge accordingly — demand day-boat product. They want shrimp that came off the boat this morning and arrives at their kitchen by two in the afternoon. That means he needs to know what is coming in before it arrives, and he needs to have a buyer for it before it hits the dock. We built him a system that starts working before the boats leave the harbor. Each captain reports his departure time, heading, and intended fishing grounds through a simple text message — because none of them will download an app and most of their phones are held together with electrical tape and optimism. The system logs the departure and cross-references the fishing grounds against three data sources: NOAA tide tables for the target area, sea surface temperature readings from the nearest buoy, and historical catch data from the same captain fishing the same grounds under similar conditions over the previous three seasons. From those inputs, the system generates a catch probability estimate — not a guarantee, because the ocean does not make guarantees — but a forecast accurate enough to begin pre-selling. By six in the morning, two hours before the first boats typically return, the system has generated a supply forecast for the day: estimated total pounds by species, estimated arrival windows for each boat, and a confidence level for each estimate. It sends that forecast to him and to his dock manager, who uses it to plan cold-storage allocation and schedule the sorting crew. Simultaneously, the system begins the sell-side process. It knows each restaurant account's typical order pattern — Husk orders eighty pounds of large white shrimp every Tuesday and Friday, The Ordinary orders sixty pounds of medium whites three days a week, a newer account on upper King Street is still ramping and averages thirty pounds twice a week. It also knows which accounts have placed standing orders versus which ones order ad hoc. For standing orders, the system confirms fulfillment against the day's forecast. If the forecast shows a short catch — and every shrimper knows the days when the water is wrong and the nets come up light — the system alerts him immediately and suggests allocation priorities: long-standing accounts first, premium accounts second, spot-market accounts last. The pricing engine was what changed the business from surviving to thriving. Before the system, he charged a flat price per pound that he adjusted maybe twice a season based on gut feel and what he heard other docks were charging. The system introduced dynamic pricing based on three variables: today's supply, today's demand, and the age of the inventory in the cooler. When the boats come in heavy — twelve hundred pounds of beautiful large whites on a Tuesday — the price drops slightly to move volume fast and prevent anything from sitting overnight. When the catch is light — four hundred pounds split across six boats because a front moved through — the price rises to reflect scarcity, and the system automatically notifies premium accounts first so they can lock in their order before the supply is spoken for. The age-based pricing was his dock manager's idea. Any shrimp still in the cooler at the end of the business day drops to a second-tier price and gets offered to accounts that are less particular about day-boat freshness — the grocery stores, the catering companies, the fried-fish restaurants that bread everything anyway. If it is still there the next morning, it drops again and goes to the box program as "yesterday's catch" at a discount that customers love because the shrimp is still excellent, just not six-hours-fresh. Since the system launched, his overnight waste has dropped from an average of ninety pounds per week to eleven. The chain-of-custody tracking solved a regulatory problem and created a marketing asset at the same time. South Carolina requires traceability documentation for commercially sold seafood — where it was caught, when, by whom, and the handling conditions from boat to buyer. He used to manage this with carbon-copy paper tickets that his dock manager filled out by hand, which were illegible half the time and lost the other half. The system now generates a digital chain-of-custody record for every lot that comes off every boat: captain name, vessel ID, catch zone, harvest time, dock arrival time, initial temperature reading, cold-storage assignment, processing timestamps for heading and sorting, outbound shipment time, delivery confirmation, and final delivery temperature. That data satisfies the regulators. But it also satisfies the chefs. Three of his top restaurant accounts now display a QR code on their menus that links to the provenance record for the shrimp on the plate that night — which boat, which waters, how many hours from ocean to kitchen. One chef told him that customers photograph the QR code more often than they photograph the food. The captain coordination was the hardest part to get right because fishermen are independent operators who have been doing things their own way for decades and do not take kindly to being managed by a computer. The system learned to work with them, not over them. It does not tell captains where to fish — they would ignore it and they should, because they know the water better than any algorithm. What it does is aggregate their collective intelligence. When Captain Marcus reports heavy catch off Morris Island and Captain James reports nothing at Capers Island, that information flows to every other captain who is still deciding where to set nets. Not as a directive. As intelligence. "Heard Morris is producing today." The captains who adopted the system's information sharing report that it cut their dead-haul days — days when they burn fuel and catch almost nothing — by about a third. The direct-to-consumer box program was a side business that the system turned into a real revenue stream. Before, he packed boxes ad hoc when he had surplus and shipped them via a regional carrier with no consistency in timing or contents. The system now manages the box program as a proper subscription: customers choose weekly or biweekly delivery, select their preferred species and size, and the system allocates their orders against the daily catch forecast. Packing instructions go to the fulfillment crew with exact weights, ice ratios, and carrier pickup windows. The system tracks delivery temperatures via the insulated packaging's built-in monitor and flags any shipment that arrived above forty degrees so he can proactively reach out to the customer before they complain. Box program revenue has tripled in seven months. His favorite feature is the simplest one. Every night at eight o'clock, the system sends him a one-page daily summary: total pounds received by species, total pounds sold, total pounds in cold storage with age flags, revenue for the day, margin for the day, and a one-line forecast for tomorrow based on which captains have confirmed they are going out. He reads it on his porch with a beer, and for the first time in nine years he does not go to bed wondering whether he forgot to call someone back or whether the walk-in is set to the right temperature. "My grandfather ran this business out of his back pocket," he said. "My father ran it out of a filing cabinet. I was running it out of my phone and losing my mind. Now I run it from a dashboard and I actually know what is happening. The shrimp are still wild, the captains are still wild, and the ocean still does whatever it wants. But the business part — the part between the dock and the plate — that part finally makes sense."