Dunkin’ Uses AI to Cut Donut Waste & Boost Profits

AI‑Powered Donut Forecasting: What Bluemont’s Do’Cast Reveals

When a fast‑food chain with nearly 100 Dunkin’ locations discards a few leftover donuts each day, the cost can snowball. Fast Company reports that the Knoxville‑based franchisee Bluemont Group was losing just under $100 per store per day in unsold treats—more than $3 million a year in waste. To tackle this, Bluemont teamed up with AI specialist PreciTaste and launched Do’Cast, an in‑store camera system that predicts donut demand in real time.

Do’Cast blends image‑recognition with predictive modelling. Cameras count each donut as it leaves the case, while the algorithm weighs recent sales, weather, holidays, day‑of‑the‑week patterns, and even local events such as college football games. The result? A waste reduction of up to 25 %, translating to “cutting that in half” and saving roughly $1.5 million annually.

Why Real‑Time Vision Beats Manual Logs

Busy Dunkin’ crews don’t have time to log every pastry, especially when customers mix and match flavors. Do’Cast’s visual AI sidesteps this bottleneck, instantly distinguishing between an Old Fashioned and a Chocolate Crème—something pre‑existing bakery AI in Japan had already proven possible. PreciTaste’s background in German‑engineered AI ovens, which automatically recognize breads, rolls, and croissants, gave the team confidence that “if they can identify breads, they can identify donuts,” says Bluemont’s Margo Hughes.

Emerging Trends Shaping the Future of Fast‑Food Forecasting

Hyper‑local demand modelling: Integrating weather, local events, and even traffic data to fine‑tune inventory at the store level.
Computer‑vision inventory audits: Using existing security cameras for real‑time stock counts without extra hardware.
Dynamic supply chain coordination: Linking store forecasts directly to central bakeries so daily deliveries match predicted demand.
Sustainability dashboards: Translating waste‑reduction percentages into carbon‑offset metrics for brand storytelling.

These trends echo the success story highlighted by Restaurant Business, which notes that AI can “count its doughnuts” and keep popular flavors on the shelf.

Did you know? A single Dunkin’ franchise can waste over $100 each day—just the cost of unsold donuts—before AI steps in.

Pro tip for operators

Start with a pilot in high‑traffic locations. Capture baseline waste data, install a camera‑based AI system, and compare weekly waste before scaling. Small adjustments in the product mix, guided by real‑time AI insights, can quickly generate “sweet spot” savings.

Frequently Asked Questions

How does Do’Cast know which donut is which?: It uses image‑recognition algorithms trained on the visual features of each donut type, similar to AI that distinguishes baked goods in Japan.
What factors influence the demand forecast?: Recent sales, weather conditions, seasonal trends, holidays, day of the week, and major local events such as college football games.
Can this technology be applied to other menu items?: Yes. PreciTaste’s origins in AI‑powered ovens that recognize rolls, breads, and croissants suggest the model can expand to broader bakery and snack categories.
What kind of cost savings are realistic?: Bluemont reports a reduction in donut waste of up to 25 %, turning a $3 million annual loss into roughly $1.5 million in savings.

Ready to explore how AI can sharpen your store’s inventory? Contact us today for a free consultation, or read more case studies on AI‑driven foodservice innovation.