Walk into a grocery store with a list of 25 items. Leave with 18. As far as the store’s data warehouse is concerned, that was an 18-item trip. The seven items you wanted but didn’t find, or got distracted from, or couldn’t justify in the moment, are invisible. The transaction log is a record of what happened. It is not a record of what you wanted to happen.

Multiply that gap across 26 billion grocery trips a year in the United States. That is what the industry calls pre-purchase intent data, and it is the holy grail of grocery commerce. It is also the part of the dataset that no retailer, no delivery platform, and no point-of-sale system has ever owned.

Stretch does. Because Stretch sits on the shopper’s side of the transaction.

What Is a Shopper-Side AI?

Walmart has Sparky. Amazon has Rufus. These are AI agents, and all of them are agents of the retailer. Their job is to optimize the basket for the store that built them.

A shopper-side AI is the opposite. It works for the shopper. It starts with the list, not the storefront. It compares what the shopper wants against what is actually in stock, at what price, at every nearby store, and returns the cheapest, closest, or fastest answer depending on what the shopper says matters today.

That distinction is the whole product. It is also the reason the data Stretch produces is genuinely new. We see what shoppers planned to buy, before any store gets a chance to convert or lose that intent.

Why This Matters Right Now

The macro backdrop is brutal. US food-at-home prices have climbed roughly 25–30% since 2020 (USDA Economic Research Service). The single sharpest year, 2022, was an 11.4% increase, the steepest US grocery inflation in more than four decades. Even with headline inflation back near 2.4%, food costs have not given the gains back; they have just stopped rising as fast (US Bureau of Labor Statistics).

Households have responded the only way they can. Circana’s 2024 outlook found Americans are making 8.9% more grocery trips a year while buying as much as 11% fewer items per trip. We are shopping more often, with smaller baskets, hunting for better prices. The behavior change has happened. The tooling has not caught up.

Against that backdrop, the most useful thing a piece of consumer technology can do is compress the distance between what shoppers planned and what is actually available nearby. That is what a shopper-side AI does. Every time it does it, it also generates the data point retailers have never had.

The Food Waste Reframe

Pre-purchase intent has an unexpected use, it is the missing piece in the fight against food waste.

ReFED estimates the United States generated 70 million tons of surplus food in 2024. Consumers alone wasted nearly 35 million tons, at a cost of about $261 billion, the majority of that on groceries bought and never eaten. The retail slice of the waste pile, the food that never even makes it off the shelf, is worth roughly $100 billion a year on its own.

Today, that food is thrown away because nobody knows it should have gone somewhere else. Picture the inverse. A store has 50 packages of strawberries close to a sell-by date. Three shoppers within two miles have strawberries on their Stretch list this week. The store offers a discount, those shoppers route to the right aisle, the food gets eaten, and a loss line becomes a revenue line. The matching is trivial, once both sides exist. The shopper-side data is the half that has not existed before.

What Comes Next

Five years from now, the goal is to have as many US shoppers as we can broadcasting their intent to a platform that represents them rather than the retailer.

The foundation is already in place. A shopper-side AI. A 500,000-product database. And a live app in the Apple App Store and on Google Play.

That’s Stretch: stretchgroceries.com.