Two AI cart builders, one week apart
Instacart launched Clementine and Shipt launched Ask Shipt within days of each other in early September, and the overlap in what they do is not a coincidence. Both take a loosely specified request such as a photo of a recipe, a handwritten list or a phrase like a Saturday tailgate for twenty five people, and turn it into a ready-to-buy cart pulled from real-time inventory and pricing at a chosen store. Both also surface deals, swap in lower-cost alternatives, and learn from a shopper's purchase history to personalize future carts.
Instacart CEO Chris Rogers described Clementine as putting nearly fifteen years of learning how families shop and eat to work, while Shipt's Katie Stratton framed Ask Shipt as bringing people-centric AI directly into the shopping journey. The near-identical framing from competing executives is the real story here: this is no longer a differentiated feature either company can claim credit for originating. It is now table stakes for any grocery marketplace competing for basket share.
The data moat behind Clementine specifically
Instacart's version leans on a scale advantage that is genuinely hard to replicate: 1.6 billion lifetime orders, a catalog of more than two billion items, and over ten million daily inventory signals pulled from nearly 100,000 stores across more than 2,200 retail banners. That depth is what lets Clementine confidently suggest a specific brand of pasta sauce is on sale at a specific nearby store right now, rather than offering a generic recipe recommendation disconnected from real inventory and price.
Instacart is also white-labeling the underlying technology as Cart Assistant, already live with Food Bazaar, Heritage Grocers Group and Woodman's, with Aldi US, Harmon's, Save Mart and Stew Leonard's confirmed to follow. That licensing move matters more than the consumer-facing Clementine launch for most of our readers, because it is how a regional grocer without the data scale to build this in-house gets access to it anyway, at the cost of routing shopper data and interface control through Instacart's stack. Shipt, by contrast, is leaning on breadth across its marketplace of more than one hundred retailers rather than a single dominant data pool, a different bet on where the value in this category actually sits.
Why both are racing into the same AI assistants
Neither company is stopping at owning its own app experience. Instacart has integrated with Anthropic, OpenAI and Google, meaning Clementine's underlying capability is becoming reachable through Claude, ChatGPT and Gemini directly. Shipt has already integrated with ChatGPT and Claude ahead of this launch. The strategic logic is straightforward: if AI-native shopping is genuinely a new discovery surface, both companies would rather be the fulfillment and inventory layer behind every major assistant than bet the entire business on their own app remaining the default entry point.
That is a defensible hedge for the marketplaces themselves. It is a harder position for the grocers and CPG brands whose products actually populate these carts, because it means the same inventory is about to become shoppable from at least four separate AI surfaces simultaneously, each with its own ranking logic for which brand or product gets suggested first when a request is ambiguous. None of those ranking systems are transparent to the brands whose products they are ranking, which leaves merchandisers optimizing for a black box on four fronts at once instead of one.
What grocers relying on these platforms are giving up
A regional grocery chain that adopts Cart Assistant or a Shipt integration gets AI-driven basket building without building it, which is a reasonable trade for a chain without the engineering budget to compete on this feature natively. But it also means the shopping interface, the personalization logic and the resulting first-party purchase data increasingly live inside Instacart's or Shipt's systems rather than the grocer's own. The chain becomes a fulfillment location inside someone else's AI shopping experience.
That is not a reason to opt out, since the alternative of building a comparable AI cart builder in-house is a multi-year, capital-intensive undertaking most regional grocers cannot justify. It is a reason to negotiate data access terms now, before the integration is load-bearing for a meaningful share of digital orders, rather than after switching costs make renegotiation impractical. A grocer that waits until AI-driven orders are twenty percent of digital volume to ask for its purchase data back will have far less leverage than one that puts the terms in writing at signing.
The decision this puts on CPG and grocery IT teams
For CPG brands, the practical implication is close to what we described for Estee Lauder's AI discoverability push: if Clementine and Ask Shipt are increasingly the interface through which a product gets chosen, product data quality directly determines whether a brand gets recommended or silently passed over in favor of a cleaner-catalogued competitor. Brands with inconsistent product titles, missing attributes or stale pricing feeds will simply be invisible to an agent optimizing for a confident, fast recommendation.
For grocery IT teams, the near-term task is auditing exactly what data these integrations expose back to the retailer, whether that data arrives in a format usable for the retailer's own demand forecasting and merchandising systems, and whether the contract terms allow the retailer to build a competing capability later without penalty. Treat this integration the way you would treat a marketplace listing agreement with Amazon, not a marketing partnership, because the leverage dynamics are the same.
What to watch next
Watch whether DoorDash, which already offers a comparable Cart Assistant-style feature at Albertsons, Kroger and other chains, responds with its own scale claims, since a three-way race on data depth will pressure all three platforms to publish more specific performance numbers than the launch announcements currently include. Also watch whether any of the major grocers with enough scale to build in-house, Kroger and Albertsons chief among them, choose to compete on this feature directly rather than license it, since that would be the clearest signal yet of how much strategic weight retailers are putting on owning the AI shopping interface themselves.
For now, the practical takeaway for retail and CPG technology leaders is that AI-driven cart building has moved from a single company's experiment to an industry baseline in the span of about a week. Any 2027 roadmap that still treats this as an optional, future capability is already behind where the market has settled, and the vendor selection question has shifted from whether to integrate to which platform's data terms a retailer can actually live with long term.
