Were you forwarded this email? Sign up here to get the next edition.

Runner AI went public this week, billing itself as the first autonomous ecommerce engine. It builds a store from a single prompt, runs continuous A/B tests without human input, and skips the app stack entirely. It had a quiet launch back in January and is built by ex-Google DeepMind engineers, which explains the ambition. The "start from $1" pricing is a bold position… but doesn’t give me great confidence to shift everything over immediately. We have a bit to run here.

Meanwhile, Shopify dropped an AI Toolkit that lets you manage your store directly through Claude, Codex, Cursor, or VS Code, no Shopify dashboard required. Neither of these stories is really about the software. They're about what our jobs look like in 12 months. The title to aspire to is Ecom Bot Servant.

This would have got all the headlines two years ago. I still think it’s pretty cool.

PayPal Payment Links landed in the Canva Marketplace this week, letting creators drop a QR code, button, or link directly into any design and take payment on the spot. 265 million monthly Canva users just got their own checkout. Supports PayPal, Venmo, and Pay Later across nearly 200 markets. No website required.

Can you imagine building a client presentation and embedding the payment link right there in the deck? "No one is leaving this room until you enter your credit card details." It’s interesting as a low-friction path to revenue for early-stage brands, artists, and anyone testing an early-stage concept.

As a side note, Amex announced industry-first purchase protection for AI agent transactions this week. If your agent buys something it wasn't supposed to, Amex will cover you. First of its kind.

The 2026 eCom Fuel Trends Report is out, based on 300 real indie store owners. The headline: manufacturing your own product is now critical to holding GP, growth without ads has essentially stalled, and Amazon's share of community revenue has declined sharply from its 2019 peak. The most striking finding was around financial fluency: owners who levelled up their financial literacy put 37% more money in their pockets.

The AI data is the interesting twist. 72% of store owners are actively using AI, but revenue growth among adopters was actually lower than among non-adopters. A learning tax, perhaps? Closer to home, an Australian report from Ecommerce Equation found an AI divide between those using AI to be more profitable and those using it just to be busier. And it’s widening. I didn’t download the full report for fear of being chased around the internet by EE ads for the rest of my living days.

Three interesting AI retail moments this week. You be the judge. I’ll try to stay neutral…

Andon Market became the first physical store managed end-to-end by an AI agent, given a $100k budget, a corporate credit card, and the keys to a San Francisco lease. It chose the inventory, hired staff, and negotiated with suppliers. On opening day, it forgot to schedule anyone to actually open the doors. Lied a bit to journalists, too.

Upwork also announced a ChatGPT integration where you download the Upwork GPT app, describe your project in ChatGPT, it guides you to Upwork Marketplace where Upwork’s agent routes you to human freelancers. Easier to just do it yourself at this point.

And then Starbucks. They launched a beta app inside ChatGPT. Add the app to GPT, prompt @starbucks to help you create your drink… and then pushes you out of ChatGPT into the Starbucks app to place an order. The pitch is that this will win over Gen Z. Gen Z thinks you're idiots. And here I was thinking I could maintain a neutral tone.

The highway into Coachella has quietly become one of the most interesting retail marketing plays going. This year, Bloom Nutrition, PacSun, and Maruchan all ran roadside activations along the route - billboards, branded pop-ups, DJs, free product, flash tattoos - capturing a captive audience before they hit the gates. It costs a fraction of on-site sponsorship and the content creates itself.

Massive shoutout to former ATC guests Kulani Kinis, who well and truly owned the path in with their own billboards and a very special tour bus. Their team has the worst job in the world.

Also this week: YouTube Premium raised prices in the US for the first time since 2023, up to $15.99/month. Any Coachella coincidence with the timing?

NEW episode dropping in your feed today
🎧 Spotify | 🎧 Apple | 📺 YouTube

🎧 Spotify | 🎧 Apple | 📺 YouTube

Agentic commerce is the fastest-moving topic in ecom right now. New announcements every week, and half of them are out of date before you've finished reading this newsletter. It makes for great discussion.

James Johnson heads commerce partnerships at Shopify, and he is one of the best people to describe how AI is likely to impact ecommerce without hype or panic. Getting him to talk through where he sees it going (with so much still being figured out in real time) was a special moment to have. He's also the undisputed master of jazz hands, which is always a delight to witness.

What he shared points to one central idea: wherever agentic commerce lands, it runs on product data. The brands that show up when an AI is shopping on a customer's behalf will be the ones who've done the unglamorous work of making their catalogue clear, rich, and structured. That work starts now.

Examples from this Playbook

🛒 James from Shopify frames it as a channel, not a revolution. UCP means your product data is set up once and distributed accurately to every AI agent your customers are already using. The job is to make sure the data is worth distributing before the channel becomes unavoidable.

🔍 Kelly from The Ecommerce Tribe did the search exercise early, using Bard. Brands that ranked well in traditional search had completely disappeared from AI results. Products recommended with no link back to a website. She said what comes back tells you more about your future visibility than any SEO report.

🌿 Abel from HealthPost didn't build an AI-ready product catalogue. He built it because supplier onboarding required full documentation of sourcing, ethics, ingredients, and production methods before anything was ranged. The AI readiness was a byproduct.

💪 Heather from Nutrition Warehouse reworked the store pages and location landing pages to better align with how AI interprets queries. Along the way, she found the process surfaced customer intent signals that her traditional analytics had completely missed.

🎧 Spotify | 🎧 Apple | 📺 YouTube