Why Perplexity forfeited its AI shopping first-mover advantage

Good morning.

A quick note that I’ll be taking time off next week for the first time since I launched The Aisle more than seven months ago. My brother and nephews are visiting from New Zealand and staying with us. I think it also happens to be a good time to refresh and reset for the second half of the year. I appreciate your understanding.

That said, I will still be sending out a newsletter, but it’ll be a best-of-The-Aisle-type post rather than a piece of original reporting. If my week off coincides with your seven-day free trial of a paid subscription, I’d be happy to extend it by another week. Just reply to this message.

Moving on…agentic commerce maximalists rejoice! Adobe said earlier this week that AI engine traffic to U.S. retail sites outside of Amazon jumped 89% year over year during Amazon’s Prime Day sales event, and converted 40% better than traffic from non-AI online sources. That data adds to existing evidence that people showing up to shop at a retailer or brand’s website via AI have often already done their homework before they click buy.

The question many online retailers, big and small, continue to ask themselves is what’s the best way to successfully convert that traffic once those customers enter your virtual doors. Luckily, there’s a whole ecosystem of startups and legacy vendors alike willing to pitch you on their proposed solutions. Enjoy!

Finally, while much of The Aisle’s coverage, including the main feature below, focuses on AI’s impact on online commerce, I also enjoy nerding out on logistics and supply chain innovations from all my years covering Amazon and Walmart’s related efforts, including in my book Winner Sells All. I also happen to think so-called “physical AI” is an area we’re going to see profound disruption over the next decade.

So I was intrigued when Samsara, the $20 billion public company behind AI dash cams and telematics sensors that track cargo and equipment across the supply chain, invited me to Las Vegas for its annual customer event. As part of a paid partnership with Samsara, I interviewed MIT supply chain research scientist Angi Acocella and Samsara VP of product and engineering David Gal for a live taping of MIT's Supply Chain Frontiers podcast. (I had full editorial control over the conversation, and Samsara did not pay for any mention here.)

And I think there were a few takeaways worth a quick mention:

Firstly, there was some real excitement at the event about a new Bluetooth-tracking label that Samsara unveiled, which pairs a thin, disposable tag with the company’s existing network of connected trucks and equipment to track shipments en route, while making invisible cargo theft harder to pull off. But Gal also conceded the tech is not ready for blanket deployment: the per-label cost still only pencils out for time-critical or high-value shipments, and while coverage is strong where Samsara’s footprint is dense (North America and parts of Europe), global shipments aren’t in the sweet spot today.

Secondly, Acocella reminded us that the gap between agentic hype and reality is wide in supply chains. Her recent research at MIT found most supply chain organizations are using AI on narrow point problems, where a human still has to ask the right question or kick off the workflow and then sanity-check the result.

Not surprisingly then, when I asked Acocella about a future where AI shopping agents might talk directly to AI supply chain agents, she pointed back to the past hype cycle that claimed blockchain was going to revolutionize supply chains as a cautionary tale.

“The issue when it comes to supply chains and transportation is that there is a physical truck that needs to show up; there’s a human that is driving that truck; there’s information that needs to be shared with them; there’s real rubber on the road,” she said. “It’s not just a digital computed type of transaction.”

Sanity checks are good. Thanks, Angi.

Now on to the good stuff…

One of my first articles ever about AI’s impact on e-commerce was published by Fortune on November 18, 2024, coincidentally exactly one year before I launched The Aisle. The piece covered a new shopping recommendation experience and a new AI-powered visual search tool from the AI search engine Perplexity, marking one of the first dedicated shopping-related feature launches among the biggest AI chat apps..

Perplexity was indeed a first mover in the space.

A few months later, in March of 2025, I reported on Perplexity’s partnership with a small e-commerce software startup called Firmly that would help the AI app more easily scale its in-app checkout capability across many merchants. In November 2025, a similar partnership with PayPal went live. And finally, in January of this year, specialty retailer Pacsun announced that Perplexity users would be able to purchase merchandise from the retailer directly within the AI search engine.

It seemed like momentum.

But since then, it’s been mostly crickets from Perplexity on the shopping front. So I reached out to a Perplexity exec to find out why.