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

Hopefully see you there! Well, not the hotel room floor. That would be weird.

Cheers
Bushy

The Australian government has put forward a vague approach to its AI policy and opened an AI Office. Sounds like most CIOs' transformation plans. Most notable for our work: Australian writers, artists and journalists retain ownership of their work, and no company can use Australian creative works to train AI without the artist's control. This will really piss Pauline off with her Robert Irwin cartoons.

It follows the European model of protecting creations and calling out AI-generated work, meaning the fight for true creativity isn't over yet. It's obviously good if the sentiment is forcing companies like Meta to stop allowing users on public accounts to modify photos using AI.

But what does that mean for brands like Dollar Shave Club that are going "all in" on generative AI to showcase men's ballsacks in different ways? Surely there are exceptions that need to be written into the legislation.

New research from Gartner shows customers are three times more likely to use a third-party genAI tool such as Claude or ChatGPT than a brand-owned chatbot for customer service. Quick, someone put that chatbot project on hold! It's not all bad news though. Brands that use generative AI to answer questions and take action remain highly valuable. Just make your chatbot answer "where is my order" and be able to change delivery addresses, and you've got yourself a winner. This is 2026 innovation.

Genesys also released a report stating 50% of consumers would rather do anything, anything, than contact customer service, 84% will give a virtual agent three attempts to resolve their issue, and 48% of companies do not pass data from virtual agents to human agents. But just quietly, what really gets my goat are technology companies that release reports and still use stock images you recognise from other reports. You know you can generate any image you want these days.

Interestingly (and when did this turn into a political newsletter?), a new US Senate bill that's been presented but not passed would allow consumers to bring their own chatbots into major platforms such as Amazon and Google to do the work, rather than relying on the retailers' chatbots. A fascinating spot to watch.

In other news, China just outlawed love affairs with chatbots. I guess you can say we're coping pretty well with our new relationship with chatbots.

Prime Day results were in, and in the US, sales were up 9%, a mildly promising signal as we prepare for peak. Interestingly, those who landed on a site from an AI source converted 40% better than non-AI channels. This could be a result of Amazon blocking ChatGPT's crawlers from accessing its listings and forcing them to visit the website. However, I'm finding this across all my client research: add-to-cart rates and volumes from AI-generated traffic are still low, but conversion rates are unimaginable. Double figures at the minimum. To the chatbot point above, if customers are falling more and more in love with their chatbots, it's not just a search engine; it's a trusted recommendation.

While the demand was huge, we're not sure the deals were that great. Visualping found almost nine in ten of the products on sale by Amazon were not new lows (Amazon called the analysis "misleading", as it would).

Also interesting from Adobe's research is that influencers converted shoppers 11 times more than social networks overall. Go make your cash now, influencers.

A quiet week on the non-AI tech, but big news from Google and Pinterest: on the 25th birthday of Google Images, Google is releasing a redesign that turns it into more of a collage of images, with the ability to save images to your personalised collections and have them intelligently tailored to your unique interests. They want to make it easy for you to explore and be inspired. They want to create Pinterest. US only at the moment.

If you can't find what you want, Google lets you describe the image you want to create right from the search bar, and using Nano Banana, you'll get it in the search results. Remember when “similar images” blew our minds? While it's not here yet, it will be interesting to see whether it dents Pinterest traffic to your site, and whether there's a clear line from Google Images into their Universal Cart.

Two months ago, Pangram, a research-first company, released a Chrome extension to help people identify AI-generated content as they browse online and on social. They've just released some data from it, and in a surprise to absolutely no one, LinkedIn is the biggest AI slop machine with 40.5% of long-form content and 24% of comments AI-generated. Unsurprisingly, Substack came out the most human of the lot. No word on how many ecommerce tragics have accidentally sent AI slop email newsletters while on holiday.

In positive news, Hacky Sacks are trending again as something AI can never replace: the ability to nail your friends at Hacky Sack. Long live Hacky Sack, the great human authenticator.

If you work in ecommerce, you don’t have to figure it out alone. Inside the Add To Cart Community, you’ll find like-minded professionals, expert insights and live webinars. All for free.

This week’s discussions include:

🛠️ The "easy in, easy out" app challenge
Off the back of Martin Cox on Monday's episode, this week's challenge is to run your Shopify app stack through his gut-check: a good app works in a day, plugs into Flow without a developer, and comes out clean if you change your mind. How many of yours would survive?

🤖 James on the new ChatGPT Ads Manager
James Hallam has clocked OpenAI quietly opening up ads.openai.com and wants to know who's actually kicked the tyres. If ChatGPT is where your customers now research, are you going to be advertising there too?

👋 Say hello to Sam Penny
New member Sam founded Safety Sam, an Aussie brand helping caravan and 4WD owners monitor tyre pressure through an app-based TPMS. Give him a warm welcome.

Not in the Add To Cart community?

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

Most ecommerce teams don't need another tool. They need to actually use the ones already on the invoice.

Pull up your software bill sometime - not the total, the line items. There's a platform you signed up for two years ago and use a third of, an app someone added for one campaign that's still billing, and a tool that does more than you've ever asked it to.

It's about to get messier. Every new AI tool overlaps with three you already run, and half of them get surpassed or folded into something else within months. The real problem usually isn't the tools. It's that we're running a fraction of what we already pay for, and none of it talks to each other.

This one comes from my conversation with Ed Hallen, co-founder of Klaviyo, when he got off stage at K:SYD. Pricing changes, attribution, the SaaSpocalypse - all the big stuff got covered. But the part that stuck was how he thinks about open platforms: the brands plugging Klaviyo into the rest of their stack, including through the MCP, end up driving more value, not less.

Examples from this Playbook

🧾 Ed from Klaviyo makes the case for open over locked-in. Brands using the MCP don't spend less time in the platform; they get more out of it.

👗 Doug from Incu put the tech and the effort straight behind the customers who matter most. Point your best tools at where the money already comes from, not the fringes.

🧩 Johan from Elite Supps runs a stack where every tool has one job and hands off to the next. Nobody has the one tool that does everything. Stop looking for it.

🔧 Jevon from Keeyu gave the honest version of what deep integration work actually costs.

🎧 Spotify | 🎧 Apple | 📺 YouTube

Hey, if you’ve read this far, let me know what you think. Just hit reply and give me your deepest, darkest thoughts. Would love to hear from you.