How Instacart Is Bringing Physical AI Into The Grocery Aisle

NK MEDIA
4 followers
3 Video Views·Jun 25, 2026
E-commerce changed how we shop online, but what happens when artificial intelligence enters the physical grocery store? David McIntosh, Instacart’s Chief Connected Store Officer, breaks down this massive shift happening in the retail industry. From smart "Caper" carts that recognize products in real-time to fixing the dreaded out-of-stock dilemma for grocers, David explains how bridging the gap between online data and in-store physical AI is unlocking a totally new shopping experience.
AI Generated·May contain errors
What does it mean in a world where consumers are planning via chat GPT, planning via Gemini,
planning via Anthropic? How is the role of the retailer going to change? I'm encouraging
retailers to take a more holistic view to the answer to that question.
I'm David McIntosh. I'm Instacart's chief connected store officer and a technology
entrepreneur. The premise behind connected store is to bring the convenience and the
personalization that our users love about the online grocery business
into the store and really unify the two together.
The big challenge with AI is how do you deliver a personalized experience for customers in the
store? How do you make recommendations right for them? And the proposition itself is really
simple and delightful. Now, to do that recommendation well requires a lot of sophisticated technology.
I think a lot of retailers today are very focused on online. They're seeing the rise
of hyperscalers and they're thinking about how do they deliver online experiences. And increasingly,
I'm also starting to hear retailers think about we need to have the best understanding
of what consumers want. Where do they pause? Where do they dwell? What do they substitute for another
item? What convinces them to pick item A over item B? How do they have the best understanding
of what's on their shelf, what's not on their shelf, what needs to be restocked? That's ultimately
the source of data that's going to give retailers the best understanding of their store and their
shelf, which is going to enable them not only to compete but to thrive in this new AI era.
We're bringing AI to the real world environment. And there's a couple different ways that we're
doing that. One is with our AI-powered smart cart called Kaper. This is a digital shopping cart.
Think like a normal shopping cart but with a digital screen. It's got camera sensors, it's got
location sensors, weights and measures certified scale, an NVIDIA Jetson board, NVIDIA GPU at every
single cart that's running the sensor fusion algorithm to understand what people are putting
in the cart. And for consumers, it gives them a running total so they can keep track of what
they're spending. We also have a product called StoreView where both Instacart shoppers that are
picking an order can scan store shelves to understand what's on it, or the side-facing
cameras of Kaper can pick up what's on the shelf as customers are pushing the carts throughout
the store. And out-of-stocks for retailers are one of the biggest unsolved industry problems.
And so by capturing pictures and imagery of the shelf and pairing that with an AI algorithm that
lives at the edge, that lives on device, we can actually start to help retailers solve the
out-of-stock problem, ensuring that customers get what they ordered online but also improving the
in-store experience.
What Kaper is doing is helping customers save more money, so customers are getting more access
to deals, to discounts, as I mentioned earlier. We're also seeing significant basket size increases.
Retailers have reported double-digit basket increases that they're achieving with Kaper
because we're able to not only deliver a seamless experience to customers in-store, but by
understanding their omnichannel history, where the cart is in the store, the location,
we're able to deliver recommendations that are fundamentally more useful to their journey.
Another thing is that we're starting to unify the online and in-store experience. So, for example,
we'll enable a customer to take a shopping list that they built online in an app and sync it to
the Kaper cart in the store, and the cart will remind them as they're shopping the store so
they don't forget anything on the list. More broadly, we're connecting all the elements of
the store together. The way we think about AI is to really enable our grocery partners in the
industry to better compete. So many of our grocery partners don't have the R&D budget
to be able to build the same technology with the quality of, say, a Google or other leading
technology companies. And so, really, the way we think about our approach to technology is to arm
our retailer partners with the best-in-class solutions. If you think about our scale,
we power hundreds of retailer storefronts. We have 1.6 billion lifetime grocery orders online.
We're now collecting millions of sensor inputs in-store every single day from our in-store
products like Kaper. And so, our approach is to really enable our retailer partners and enable
them to benefit from the scale of data that we're capturing online and in-store, and really enable
them to benefit from the technology that we're developing to power their storefronts as well as
their experiences in-store. One of the mistakes that we've seen organizations make when adopting
new technologies like AI is when they don't think about the full ecosystem they need to activate
to really get a technology to be used. One of the things that we've seen essential in driving
adoption is ensuring that associates are on board with the technology. Without associate
support, without the employees really getting behind it and putting their shoulder in it,
it will not be successful. And so, that's been one of the keys to our success, which is we
fundamentally design products. We think about the entire ecosystem. We think about the retailer
and the user and the CPG and the associate. Everything we build ultimately has to work for
all of those four stakeholders. Otherwise, it's simply not going to get adoption in the industry.
