The AI Tollbooth Is Already Charging Rent

Hi there,

I've been talking about the AI tollbooth for months, and this past week made it feel more concrete than ever.

Google launched its Universal Cart. Shoppers can now discover and add items from Nike, Sephora, Target, Walmart, and Wayfair into one cart across Search, Gemini, Gmail, and YouTube [1]. A Wharton professor studying this shift said it plainly: "Whoever controls the agents now has the power."

That's the tollbooth. It's not one platform; it's a layer every major AI surface is building between your brand and your buyer. And every time a new one appears, the cost of not having identity infrastructure goes up.

The tolls are compounding

ChatGPT's ad program crossed $100M in annualized revenue within six weeks of US launch [2]. Google Universal Cart is live. TikTok Shop is on pace for $20 billion in US sales this year. Shoptalk 2026 made agentic commerce the centerpiece.

Each of these is a tollbooth with its own algorithm, its own data, and its own signal about what gets surfaced. You can show up inside each one. But you're renting that audience, not owning it.

Someone sees your ad in ChatGPT and buys offline -- you can't attribute it, so your ad platform never learns to find more people like them. Someone buys you on TikTok Shop, TikTok keeps the buyer data, not you. Someone adds you to Google's Universal Cart, you have no way to identify them, add them to a retention flow, or suppress them from future spend once they've already bought. Every conversion that happens inside the tollbooth disappears into it.

The brands getting this right started earlier than you think

81% of retail executives believe generative AI will weaken brand loyalty by 2027 [3]. I think they're right -- for the brands that keep renting the audience without ever owning the relationship. The ones that prove them wrong are the brands that built direct customer identity before the tollbooth appeared.

What that means in practice: knowing who your retail buyers are and wiring that signal back into your ad platforms (Meta, Google, TikTok) so that when someone walks through the AI layer and arrives at your brand, the algorithm already knows who they are. And who to find more of. The tollbooth stops being a cost and starts being an advantage.

As each new AI layer appears, the brands that haven't closed this loop pay every toll from scratch. The brands that have see their signal quality compound. Not because they're spending more, but because the algorithm keeps getting smarter about their customers when brands feed them the right signal.

Before the next one appears

Every quarter, another AI surface enters the funnel. The question to bring to your paid media team: when a buyer discovers you through any of these platforms and converts -- online, in-store, or at retail -- do you know who they are? Can you feed that signal back so the algorithm finds more people like them?

If the answer is no, you're paying every toll from scratch.

Talk soon,

Kait

A few extras:

On June 18th, we're sharing the exact playbook TUSHY used to identify 15K+ retail buyers from Amazon and Walmart and convert them at a 33% email opt-in rate.

Brij is teaming up with our friends at PostPilot, Klaviyo, team TUSHY, and more for The Year of the Omnichannel Moat, a free half-day workshop.

You'll hear from real operators at TUSHY, Heath-Ade, Criquet Shirts, and Mizzen+Main on acquisition, retargeting, and retention, plus tactical frameworks from PostPilot, Klaviyo, Digioh, and OuterSignal on how to stack a proven program before BFCM 2026 hits.

Can't make it? Register anyway to get the recording afterward.

Some of my favorite brands are hiring…

1. Jones Road Beauty - Vice President of Marketing

Sources:

[1] Retail Dive, "Google launches cross-retailer Universal Cart," May 2026. https://www.retaildive.com/news/google-launches-cross-retailer-universal-cart/820957/

[2] Digiday, "OpenAI's ad pilot hits $100M on marketers' fear of missing out," April 2026. https://digiday.com/marketing/marketers-join-openais-ad-pilot-nudged-by-fomo/

[3] Retail Dive, "Retail's risky AI commerce bet," 2026. https://www.retaildive.com/news/retails-ai-bet-risks-data-market-share/810202/