Starbucks this week launched a new internal tool that helps store managers automate inventory management using a combination of computer vision and AI (visual AI). Known as AI-powered automated counting (it should probably come up with a snazier name), it lets users scan inventory to quickly count it.
The way this works is that Starbucks’ personnel scan a shelf or fridge with an iPad to log all items 8x faster than traditional/manual inventory counts. Advantages include time savings and lessening cognitive load – both of which let store associates and managers divert time and bandwidth to other tasks.
Beyond operational efficiencies, this automated and digitized approach is more effective as it cuts human error out of the loop, such as faulty item counts. Altogether, the result is tighter inventory systems, which in turn engender healthier cash flow (a la just-in-time inventory) and customer service.
That last part translates to fewer situations where something is out of stock. That includes everything from CPG items that Starbucks sells, to the caramel drizzles and soy milk that go into its drinks. Given a faster and easier way to take inventory, data collection will be more frequent and up-to-date.
The technology is rolling out across Starbucks’ North America locations, installed on tablets that are already present in those stores – meaning a streamlined rollout involving bits rather than atoms. It’s already active in “thousands” of locations with a commitment to full rollout by the end of the month.
Back to Fundamentals
Panning back, Starbucks’ integration broadly aligns with an underrated segment of AI that we’ve been tracking. Though generative AI and LLM-based chat engines get most of the attention and excitement, this flavor of AI – which we call visual AI – extends to the broader reaches of the physical world.
In other words, most AI that we discuss and use is confined to the web. That’s of course valuable and expansive, but relatively confining considering the volume of commerce that happens offline. As it’s often stated in the halls of Localogy, meatspace commerce outnumbers e-commerce by about 9 to 1.
Other examples of visual AI include Augmodo. As we recently examined, using simple cameras affixed to store-associate lanyards, it passively gathers spatial maps of store shelves as those associates wander aisles in the regular course of their jobs. Like Starbucks, one outcome is inventory optimization.
But beyond visual AI’s value in operating in the physical realm, it carries another key AI attribute: saving local businesses time via automation. This continues to be at the top of the list of success factors for AI integrations. It’s also a data-backed notion, as it was validated in a recent SMB survey we did with Duda.
The kicker is that this isn’t anything new. Though AI is framed as a new and buzzy thing, its most valuable integrations tap into fundamentals: saving time and headaches. And the active ingredient in that formula isn’t necessarily AI, but another “A” word: automation. We’re now just getting better flavors of it.
