Morning All,

No preamble this week, let's get straight into it. Your product could literally be the best thing since sliced bread, but if nobody knows about it, or no-one can find it, it might as well not exist. The rules of the marketing game are changing quickly and that's what we're discussing today:

• Marketing for an AI agent audience
• Pinterest is now prime digital real estate
• Canva and Anthropic put a marketing team in your pocket

Let's discuss...

When Your Customer Is a Bot, Your Marketing Playbook Needs Rewriting

A growing share of online shoppers aren't human, and the persuasion tactics that have driven e-commerce conversions for decades simply don't work on them. New research published in Harvard Business Review tested thousands of simulated shopping rounds across four leading AI models, and the findings should make every e-commerce marketer sit up straight.

The research tested classic e-commerce persuasion tactics, you know...the kind baked into virtually every online store: scarcity messaging, countdown timers, strike-through pricing, voucher codes, and product bundles. These tools were built to exploit human psychology. Urgency, loss aversion, the thrill of a deal. Slight problem...AI agents don't have psychology. The rules of agentic commerce will be governed by logic.

When researchers tested eight common e-commerce promotional mechanisms, across four AI models, in thousands of simulated shopping rounds, they found that only two behaved consistently the way you would expect them to for human buyers.

The researchers tested:

• Assurance signals (“Money-back guarantee”)
• Countdown timers
• Strike-through pricing
• Scarcity cues (“Only 2 left!”)
• Social proof (purchase counts)
• Vouchers
• Bundles
• Star ratings.

Only ratings consistently pushed choices upward across all models and product categories that were tested. This mirrors the well-established human reliance on quality signals. Social proof was the next most impactful signal, but even that varied across cases. Conversely, higher prices consistently reduced selection.

Everything else? Unstable, model-specific, and sometimes counterproductive.

Perhaps the most striking finding is what happened with more advanced reasoning models. Rather than being nudged by overt persuasion tactics, they appeared to be actively sceptical of them. Think about that for a moment...the "smarter" the AI, the more your flashy "LIMITED TIME OFFER" banner might actually hurt you.

OpenAI has pushed ChatGPT deeper into product discovery, Google has launched a Universal Commerce Protocol to let AI agents transact across retailers, and Amazon has released tools that let its agents shop competing sites on customers' behalf. This isn't a niche endeavour anymore. AI agents are quickly becoming a meaningful customer segment.

What Should Marketers Do About It?

Master the fundamentals.

Before investing in agent-specific tactics, brands should ensure their fundamentals are airtight: competitive pricing and strong authentic reviews from many trusted platforms.

Segmentation now applies to AI models.

Marketers have spent decades segmenting human buyers by demographics, geographics, psychographics, and behaviour. They now need to treat the AI model itself as a new segmentation variable

Adapt what is presented based on who/what is looking.

If each model responds differently, the logical next step is to serve different versions of your product information depending on which agent is interacting with your site or data feed. Dynamically serving content like this is difficult today, but not impossible. Social media algorithms are proof that we can all be on the same site and not have the same experience. This is the logical direction of travel for all sites.

Understand the prompt, not just the agent.

“Find me the best-reviewed wireless headphones under £100” is a very different query to “get me the cheapest wireless headphones that ship tomorrow.” The AI model will behave accordingly. Understanding the most common prompt structures in a brands niche is a new form of consumer research that is growing in importance by the day.

This is an ongoing strategy and not a one-off

The way models respond will not be the same forever. Every major release or update can shift how an agent responds. Brands should be building simulation environments where they can run AI agents against their product pages across models, categories, and promotional configurations.

Pinterest has become the overlooked AI marketing engine of 2026

Personally, I haven't used Pinterest regularly in years, my use of it went the same way as my use of Tumblr. However it seems that Pinterest is no longer just a mood board app. In 2026, it is becoming one of the most useful places for AI-assisted marketers. Seemingly there aren't many better places to turn visual ideas into search traffic, shopping intent and long-lived distribution. Who would've thought it...and if distribution is the new moat of the AI economy, Pinterest is looking like very valuable real estate.

Pinterest isn't like another social feed. TikTok, Instagram Reels and YouTube Shorts reward speed, personality and constant novelty. Pinterest rewards planning. On Google, established domains often dominate. On Amazon, the competition is usually closest to the transaction. On Instagram and TikTok, content often disappears before it has time to become an asset. In contrast, content on Pinterest is discovered through search, saves and recommendations long after the first post goes live. The nature of the platform represents a huge opportunity for marketers in 2026. That opportunity stems from the fact that Pinterest users arrive with intent, but are still undecided. The platform is a large visual search engine with shoppers already telling it what they want...they're just not sure about who they want it from.

According to Pinterest’s own data, 96% of top searches are unbranded, which means users are often open to discovering a product or brand they have never heard of. That is uncommon for human product discovery, but not dissimilar to the dynamic brands will experience with AI agents. Learning and mastering what drives buying decisions on a platform like Pinterest will put brands in a great position to capitalise on agentic commerce. Current ecommerce systems are great at processing transactions but often suck at guiding customers' discovery processes. Optimising customer journeys through real-time understanding of context, intent, and preferences is where agentic commerce really stands out, and it looks like Pinterest is in a unique position to capitalise.

The practical move is to stop thinking in terms of single posts and start thinking in libraries. This approach not only boosts customer engagement but also reduces the need for expensive reacquisition efforts. A pin that shows ten variations of a product being used, has a better chance than a single generic product image with a vague caption. This is where AI helps the most. Not as a shortcut or excuse to produce spammy content, but as a way to generate structured creative around real search behaviour.

The title format matters. So does the image ratio, the landing page, the board context and the match between what the pin promises and the destination. The relationship between all of these elements either builds or erodes trust which in turn will not only impact human conversions but also AI recommendations. The lesson here for any business trying to get more from AI content in 2026 is use it to produce more searchable assets tied to high-intent moments. This makes Pinterest one of the more underpriced marketing channels for brands that can afford to be patient.

Canva and Claude Are Building a Marketing Team in Your Pocket

Ok so you might need to start creating more content for Pinterest. How the hell are you going to do that? Most small businesses can't afford a full blown marketing team or agency. So what do you do? Well, Canva and Claude have got you covered. What used to require a designer, a copywriter, and a project manager can now be kicked off with a single conversation.

Canva and Anthropic have just announced a partnership that drops a fully capable AI marketing assistant directly into the hands of small business owners. Through Anthropic's new Claude for Small Business solution, Canva acts as the creative layer, turning a simple marketing brief into polished, on-brand campaign assets that owners can edit, publish, and scale. Small businesses that can't afford a marketing team...now they can.

Claude for Small Business is essentially a set of connectors linking Claude's reasoning capabilities to the tools small business owners already rely on day-to-day. Canva is positioned as the creative engine within that ecosystem, handling the visual output once Claude has processed the brief.

The workflow is straightforward. A business owner describes what they need, a campaign, a promotion, a product launch, and the integration generates fully editable, on-brand design assets. No design experience required. No briefing a freelancer. No waiting.

"On-brand" is doing a lot of heavy lifting here, and it matters. One of the biggest frustrations for small businesses using generic AI tools is the output looking...well, generic. The Canva integration is built to respect brand identity, meaning the assets it produces should actually look like they belong to your business.

Canva already calls itself the world's leading all-in-one visual communication platform, with a user base heavily weighted towards small businesses and solo operators. Pairing that reach with Claude's conversational AI creates a genuinely accessible entry point for owners who've been curious about AI but unsure where to start.

If you remember nothing else: Like Thanos, a future with agentic commerce is inevitable. The fundamentals of good business, competitive pricing and genuine customer reviews, turn out to be the only marketing levers that consistently work on AI agents. The tactics we layered on top of those fundamentals, the urgency triggers and psychological nudges, were always a bit of a shortcut. Now, with AI agents making more purchasing decisions, that shortcut is becoming less effective. Brands that win the next decade of e-commerce won't be the cleverest at manipulation. They'll be the best at being genuinely worth choosing.