Standard AI

Here is a fact about physical retail that should bother more people than it does: the store you walk into is, informationally, a black box. Online, every hover and scroll is logged, joined, and modeled to death. In a store, the shelf has no analytics tab. The endcap is a guess. The $4 display that a brand paid a slotting fee for either worked or it didn't, and mostly nobody knows which. Roughly 80 percent of retail happens in these black boxes, and until recently the industry mostly measured the one moment when a black box briefly told the truth - the checkout scan.

Standard AI's entire existence is an argument that this is silly, because the store is already full of sensors. They're called cameras, and they've been bolted to the ceiling for decades to catch shoplifters. The cameras see everything. Nobody was reading them.

The company started in 2015 as a prototype in a Texas garage - a handful of engineers who had, of all things, been building systems at the SEC, the financial regulator, and decided the more interesting hard problem was teaching a computer to watch people shop. In 2017 it incorporated in San Francisco as Standard Cognition, went through Y Combinator's summer batch, and picked the most futuristic possible version of the idea to sell: autonomous checkout. Grab what you want, walk out, get charged automatically. No cashier, no line, no scanning. It is a genuinely delightful demo and it was, for a few years, the whole pitch.

Investors liked the demo. Standard raised a $40M Series A, then a $35M Series B in mid-2019 that more than doubled its valuation to $535M in about eight months, then in February 2021 a $150M Series C led by SoftBank's Vision Fund 2 that valued it at a clean $1 billion. Along the way it did the thing that well-funded startups do when they're in a race: it bought its competitors and their engineering teams. DeepMagic, a New York retail-automation shop, in 2019. Milan-based Checkout Technologies in 2020, which conveniently came with a European engineering staff. The strategy, in retrospect, was less "acquire revenue" and more "assemble a toolbox faster than you could build one."

The Pivot Nobody Announced Loudly

And then the interesting thing happened, which is that the futuristic pitch quietly stopped being the business. Autonomous checkout is hard, capital-intensive, and requires either building your own stores or convincing big retailers to re-rig theirs. Meanwhile, the byproduct of all that computer vision - a precise, continuous record of where shoppers went, what they picked up, what they ignored, how long they lingered - turned out to be something retailers and consumer-goods brands would pay for without ripping out a single register.

So Standard Cognition became Standard AI, and in March 2024 it debuted VISION, a "vision analytics" platform. The framing shifted from "we'll replace your checkout" to "we'll read your cameras." VISION measures foot traffic and impressions, maps how people move through a floor plan, and produces something the company calls a Visual Engagement Score - a real-time number for how much attention a product, a promotion, or an in-store ad actually got. It is, essentially, web analytics for the aisle.

This is a smarter position than it first appears. Retail media - brands paying to advertise inside stores and apps - is one of the fastest-growing ad categories, and it has an awkward, unanswered question at its center: did anyone actually see the in-store ad? Digital advertising sells on measurable impressions. Physical retail media has mostly sold on faith. Standard AI is building itself into the company that measures the impression. The unglamorous truth of a lot of good businesses is that they don't sell the magic - they sell the yardstick that tells you whether the magic worked.

Watching Without Watching You

There is an obvious objection to all of this, which is: cameras watching shoppers sounds like surveillance. Standard AI's answer is that the system is privacy-first by design - it analyzes anonymized movement and engagement rather than identifying individuals by face or name. This is not only a compliance posture; it's what makes the product deployable at all. A retailer can roll it across hundreds of stores precisely because the interesting signal - how a crowd behaves - doesn't require knowing who any single person is. The constraint is the feature.

The leadership reshuffle told the same story as the product. Co-founder Jordan Fisher, the CEO who pitched the autonomous-checkout dream, moved to board chairman. Angie Westbrock, a longtime retail, CPG, and technology operator, took over as CEO; David Woollard, a computer-science PhD with time at Samsung and NASA, became CTO. Vision - the human kind - starts companies. Operators scale the boring, valuable version.

A Business Built By Shopping List

It is worth pausing on the acquisition habit, because six deals in seven years is a lot for a company this size, and it says something about how Standard AI thinks. Most startups treat M&A as a milestone - a press release, a logo on a slide. Standard AI treats it more like a shopping list. DeepMagic added retail-automation know-how. Checkout Technologies added a European engineering bench. Each deal filled a specific gap in the stack rather than buying growth for its own sake. When you're racing to build a hard technical platform and the clock is a venture fund's fund life, buying the missing piece is often cheaper than waiting to grow it. The risk, of course, is that a company assembled from parts can feel like a company assembled from parts. Standard AI's counter is that computer vision is modular by nature - detection, tracking, spatial reasoning, engagement scoring - and modules can be bolted together if you own the data layer they all feed.

The numbers around the company are, characteristically for a private startup, a little fuzzy at the edges. Public estimates put revenue in the neighborhood of $25 million and headcount near a hundred - modest against a $1 billion valuation set at the top of the 2021 capital market. That gap is the honest tension in the story. A unicorn valuation is a promise about the future, and the future Standard AI is now selling - retail analytics, engagement measurement, spatial intelligence - is a slower, more grinding business than the frictionless-store fantasy that earned the valuation. The pivot is a bet that the grind is real and the fantasy was not. It is a bet a lot of former "future of retail" companies did not survive long enough to make.

The Latest Piece

In January 2026, Standard AI made its sixth acquisition: Pathr.ai, a spatial-intelligence company whose specialty is analyzing how people move through a space. Founder George Shaw joined as an SVP of AI Strategy. On paper it's a modest tuck-in. Strategically it's the missing verb. Traffic counts tell you how many people came in; conversion tells you how many bought; path tells you the story in between - the route, the hesitation, the aisle that everyone walks past. Standard AI could count and it could convert. Now it can narrate the walk.

The ecosystem around the product tells you where it's aimed. Standard AI leans on Google Cloud for infrastructure, NVIDIA for the GPU and edge-AI horsepower that vision models demand, and camera partners like Axis for the hardware already hanging in stores. That last detail matters: the pitch to a retailer is not "install our exotic ceiling rig" but "we can often work with the cameras you have." Lowering the cost of a yes is how analytics software actually spreads across a fleet of hundreds of locations, and it's the quiet difference between a science project and a platform. The company says it operates across roughly two dozen countries, which for a firm of its size implies it is selling into large multi-store chains rather than winning one boutique at a time.

What you can actually do with Standard AI, if you run stores or make the products in them, is stop guessing. Where should the display go? Which endcap earns its rent? Is the new planogram getting looked at or ignored? Did the promotion move attention or just move inventory into a corner? These are questions retailers have answered for a century with intuition and quarterly hunches. Standard AI's bet is that the cameras already know, and that reading them is worth more than the futuristic store everyone thought it was building. The garage prototype promised a world with no cashiers. The company that grew out of it sells something less cinematic and probably more durable: a store that can finally tell you what happened inside it.