The arithmetic of a grocery shelf is cruelly simple. Order too few strawberries and a shopper leaves annoyed. Order too many and the berries soften, leak and become a loss with a compostable aroma. For decades, the person writing that order has balanced sales history, weather, promotions, display size, delivery schedules and instinct - sometimes with a scanner, sometimes with a pencil, often with software designed for products that do not rot.
Afresh was built in that mismatch. Founded in San Francisco in 2017 by Matt Schwartz, Nathan Fenner and machine-learning researcher Volodymyr Kuleshov, the company makes enterprise software for grocery retailers. Its models forecast demand, estimate what inventory is probably on hand, and recommend what stores and distribution centers should buy, order or produce. The interface is meant to give the operator a useful first draft, not a lecture from a black box.
That modest distinction has become a large business. Afresh says its software now supports more than 12,500 departments across 40 states. Customers and partners named publicly include Albertsons Companies, Meijer, Wakefern, Stater Bros., Fresh Thyme, Heinen's, Grocery Outlet and Balls Food Stores. The company has raised about $181.8 million, including a $115 million Series B in 2022 and $34 million in April 2026.
The fiction in the backroom
Traditional perpetual inventory begins with what arrived, subtracts what sold and assumes the remainder still exists. In fresh food, that remainder is frequently imaginary. A cashier keys in the wrong produce code. A melon bruises. Herbs lose water weight. A customer changes her mind in another aisle. A prepared chicken becomes chicken salad, but the system records the ingredient and the finished dish as strangers.
Afresh calls its alternative Intelligent Inventory. Rather than demanding a perfect count that the physical world cannot provide, it estimates the likely on-hand quantity from item behavior, store patterns and perishability, then requests a human count when the uncertainty matters. Its forecasting models consider a range of outcomes instead of collapsing tomorrow into one confident number. The recommended order weighs availability, waste, margin and labor together.
The advantage is specificity. Bulk produce may have no barcode. Meat arrives as a random-weight case and leaves as several cuts. Bakery and deli teams create inventory during the day, often from ingredients shared across recipes. A supermarket system that treats those cases as awkward variations of canned soup produces awkward answers. Afresh built its data models and workflows around the variations.
“Fresh is different - it demands purpose-built technology.”Matt Schwartz, co-founder and CEO
A product that starts before breakfast
At store level, Afresh writes suggested orders for each item and lets associates inspect, count and override. The company reports that ordering takes 40 to 50 percent less time, while store teams follow more than 95 percent of recommendations. That second number matters. A statistically clever answer is worthless when the person facing a half-empty apple display does not trust it.
The product line has widened around the same daily decisions. Period Ending Inventory helps multiple employees count at once and flags suspicious entries. Production Planning tells a deli, bakery or prepared-foods team what to make and when. Recipe Management connects a finished item to the ingredients consumed. A manager can plan how many chickens to roast, how many to sell whole, and how many will become salad without maintaining separate realities in three spreadsheets.
Store Ordering
AI-written orders adapted to demand, shelf life, case packs and imperfect counts.
Production Planning
Item and batch plans for deli, bakery, meat and prepared foods teams.
Inventory + Recipes
Guided counts and the hidden ingredient relationships behind what stores make.
DC Buying
Purchase orders, vendor comparisons and issue resolution for distribution buyers.
At the other end of the truck route, DC Buying helps distribution-center buyers prioritize exceptions, draft purchase orders, compare vendors and respond to disruptions. Wakefern - the retailer-owned cooperative behind ShopRite and other banners - was announced as an early adopter in 2025. The idea is to connect what stores will need with what warehouses should buy, before a late crop or a promotion turns into empty shelves.
Where waste and margin agree
Afresh is a climate company by consequence and an enterprise software company by invoice. It sells B2B software to retailers, cooperatives, wholesalers and self-distributed grocers. Prices are private. Customers can deploy Afresh's mobile and web applications or feed its replenishment intelligence into existing systems through an API.
The return-on-investment pitch is unusually aligned with the mission. Less over-ordering reduces shrink, the grocery term that includes waste and other inventory losses. Better availability can raise sales. Faster ordering frees employees for work on the floor. Fresher purchases last longer after shoppers take them home. Afresh reports average results of 25 percent lower shrink, 3 percent higher sales and 7 percent more inventory turns. Those are company-wide averages, not guarantees, but individual cases offer texture: Heinen's reported $1 million in annualized savings, a 70-basis-point shrink reduction and a 1.4 percent sales increase.
The environmental claim follows the same avoided order. By the end of 2025, Afresh said it had prevented 200 million pounds of food waste, avoided 110 million kilograms of carbon-dioxide emissions and saved 4.06 billion gallons of water. The numbers are modeled impact figures, but the mechanism is concrete: food that is never needlessly ordered does not need to be grown, cooled, hauled and discarded.
From lettuce to shampoo
In March 2026, Afresh made its strategic turn explicit. The platform expanded beyond the fresh perimeter into center store, frozen goods, health and beauty, and general merchandise. It had started with the hardest inventory problem, the company argued, and could now apply that intelligence to easier ones. Grocery Outlet then selected Afresh for full-store ordering at more than 500 independently operated locations. Balls Food Stores followed with production planning and ordering across fresh departments, grocery, dairy, frozen, alcohol and general merchandise.
Broad, packaged-first
Fresh-first, now full-store
Narrow workflow
This puts Afresh into direct competition with larger retail-planning and replenishment vendors such as Blue Yonder, RELEX Solutions, Oracle Retail, SAP and SymphonyAI, plus grocery specialists including Invafresh and Shelf Engine. Its argument is architectural: rivals commonly built for packaged goods and adapted toward fresh, while Afresh moved in the opposite direction. Whether that origin remains a durable advantage as the company enters ordinary aisles will be the important test.
Expansion also changes the risk. A produce pilot can prove itself department by department; a grocery-wide decision layer touches more systems, more employees and more categories. Afresh says typical deployments finish in under four months and can sit behind a retailer's existing interface, avoiding the long replacement cycle associated with enterprise software. Grocery Outlet's rotating assortment and independent-operator model will be an especially revealing case: the platform has to respect local autonomy while finding patterns at chain scale.
The operator stays in the picture
Afresh's culture grew around store observation. The founders describe thousands of hours shadowing department workers before building the first product. The company lists kindness, candor, proactivity and humility among its values; new hires have historically introduced themselves at an all-hands meeting by naming a favorite piece of produce. It is a charming ritual, but also a useful reminder that the unit of work is not merely a SKU. It is food somebody intends to eat.
That grounding shapes the product's stance on automation. Afresh talks about removing routine execution while preserving human judgment for exceptions, supplier relationships and merchandising choices. A veteran produce manager knows that the first warm weekend changes watermelon demand; the software can bring the forecast, current inventory and margin trade-offs into one place. The manager still owns the aisle.
The useful AI is not the one with the most confident forecast. It is the one a busy department manager will check, trust and correct.
Afresh sits at an attractive intersection: enterprise SaaS, supply-chain software, retail operations and climate technology. Its customers do not need to purchase sustainability as an abstract virtue. They can buy margin improvement and receive waste reduction in the same transaction. Its shoppers may never know the name Afresh, but they can notice the result - a full berry display, fewer tired greens and a roast chicken ready at dinner.
The company's grand vision is a world where everyone has the food they need and none of it goes to waste. Grocery math will not become that tidy. Weather changes, trucks run late and strawberries remain stubbornly mortal. Afresh's more credible promise is smaller: make the next order with a clearer view of reality. Repeated across 12,500 departments, small is not a small place to start.