8 unresolved questions defining this summer of agentic commerce

Good evening.

I promise that most Knicks mentions in this newsletter will end soon. But I was so touched by the feedback to my intro in last week’s newsletter that I’ll just share a brief update.

I made the crazy call this past holiday weekend, as ticket prices in Cleveland plummeted, to drive about 1,000 miles between Monday and Tuesday with my middle schooler son to watch our New York Knickerbockers continue their improbable playoff run. (It helped that we had a free place to stay in Ohio.) And it was worth it; we witnessed our team vanquish the Cavaliers in their home arena, and earn a spot in the NBA Finals for the first time since I was in high school.

Life is short and unpredictable. Sometimes it feels really good to just do the crazy thing if you can pull it off. I feel lucky to have given my son and myself an experience I hope we both carry with us for a long time.

Now on to the good stuff that you subscribed for…

Coming off a holiday weekend and heading into summer, I figure it is worth airing some things that have stuck with me from my last month or so of industry convos. The people I've been talking to include founders, retail and AI execs, brand operators and investors, across private dinners, one-on-one interviews, panels I've moderated, and long text threads that probably should have just been phone calls.

All are trying to make sense of where AI-enabled shopping is actually headed over the coming months. Some are obviously more confident that they are the ones who will help produce the answers.

Still, a few big questions seem figured out, at least for now. The Universal Commerce Protocol (UCP), co-developed by Google and Shopify and now backed by Amazon, Meta, Microsoft, Salesforce, Stripe, and the largest retailers in the world, sure is looking more and more like the de facto standard for AI-powered commerce.

And the question underneath all of these—whether consumers will ever check out in mass numbers on non-commerce platforms— may already be getting answered, as native checkout on AI properties is mostly losing out to handoff-to-retailer models, even as Google is testing out both approaches.

Here are eight of the harder questions the industry is currently facing:

1. Are most shopping sites built on the wrong assumptions?

Spangle founder Maju Kuruvilla and Fermat CEO Rishabh Jain are among the entrepreneurs trying to reimagine what a shopping site looks like and how it works in the AI era. Kuruvilla says the share of e-commerce traffic arriving sideways—direct to a product page from AI engines, Google, social media, or elsewhere—is around 40 percent, and he’s betting that number is bound to only grow. Yet product pages are often designed for customers who enter through the online store’s front door.

Jain's version of the same type of problem: the industry has spent years chasing the promise of personalizing shopping experiences to each individual user, but based off of a static, historical profile. Instead, great shopping experiences in the AI era need to adapt to what a potential customer is doing now, in this very moment.

Yes, it's a convenient diagnosis for two founders selling brands and retailers on solutions to the problem, but the underlying argument is hard to dismiss.