B2C → B2B → B2A

Your Next Customer Isn’t Human

Commerce is quietly shifting from persuading people to being understood by machines. Meet the mistrained buyer — and find out tonight whether you already have one.

A few weeks ago, over dinner, an entrepreneur who is building his entire company on AI dragged me down a rabbit hole that kept me awake that night.

The question was simple. If people stop searching and start asking, who is actually making the buying decision?

I spent two decades in enterprise sales before starting to build software, selling industrial technology across Europe and the US. I learned the old catechism by heart: B2C means you persuade a consumer, B2B means you persuade a committee. Both assume the same thing. On the other side of the transaction, there is a human being reading your website, comparing your claims, clicking your link.

That assumption is quietly breaking. We are entering the era of B2A. Business to Agent. And right behind it, A2A. Agent to Agent, where software negotiates with software and no human reads anything at all.

Ask yourself one honest question

When was the last time you researched a purchase by typing keywords into a search box and opening ten blue links?

Now compare that with how often you ask ChatGPT, Claude, Gemini or Perplexity for a recommendation. “What’s the best CRM for a five-person agency?” “Which board game should I buy for my nephew?” “Compare these three suppliers for me.”

You are not alone. 94% of B2B buyers now use generative AI tools during their purchase process, according to 6sense’s 2025 Buyer Experience Report. Gartner predicted back in 2024 that traditional search engine volume would fall 25% by 2026 as AI chatbots absorb the queries. And during the 2025 holiday season, Adobe measured AI-referred traffic to US retail sites growing 693% year over year, with those visitors converting 31% better than everyone else.

Let me be precise here, because precision matters more than drama. Google is not dead. AI referrals are still a small share of total web traffic, just over 1% of total web traffic, growing roughly 1% month over month, according to Conductor’s 2026 AEO/GEO Benchmarks Report. If someone tells you SEO is finished, they are selling something.

But look at where the quality went. Multiple 2026 studies show AI-referred visitors converting several times better than organic search traffic, because they arrive pre-qualified. The AI already did the comparison. The human already trusts the answer. The click, when it happens at all, is a formality.

Which brings us to the uncomfortable part: increasingly, the click doesn’t happen. In Google’s new AI Mode, Semrush measured 92 to 94% of sessions ending without a single click to any website. The recommendation is the customer journey.

What B2A actually means

In a B2A world, your first customer is not a person. It is a model.

Before any human sees your brand, an AI engine has already decided whether you exist, what you do, whether you’re credible, and whether you belong in the answer. It made that decision by reading Wikipedia, Wikidata, retailer listings, review platforms, Reddit threads and news coverage. Not your homepage. Research on AI citations shows brands are far more likely to be cited through third-party sources than through their own domains.

This is the part that shocked me when I started digging. The persuasion surface has moved. For thirty years, marketing meant controlling your owned channels and buying attention on rented ones. Now the decisive representation of your brand lives in places you never optimized, aggregated by a system you cannot call.

A discipline has formed around this almost overnight. It’s called GEO, Generative Engine Optimization, the direct successor to SEO. A wave of companies now helps brands and agencies measure and influence how AI engines represent them: Profound raised significant venture money in the US for exactly this; in Europe, platforms like Aeloria are doing the same for brands and agencies who realize the AI’s answer about them is now their storefront; incumbents like Semrush have bolted GEO modules onto their SEO suites. The category didn’t exist three years ago. Now it’s a race.

Yes, you’re reading this on the blog of a company that builds one of these platforms. Discount accordingly. The shift doesn’t care either way.

The mistrained buyer

Here is the concept that kept me awake after that dinner, and it needs a name.

When a human salesperson learns your product wrong, you correct them in one meeting. When a market analyst gets your positioning wrong, you send a briefing. A mistrained buyer is something new in the history of commerce.

A mistrained buyer doesn’t dislike you. That would be manageable. It misunderstands you, which is worse, because every answer it gives about your category is built on the wrong foundation. It attributes your flagship product to someone else. It recommends you for the thing you stopped doing in 2022. Or it simply omits you, and in an answer engine, omission is not page two. Omission is nonexistence.

We met our first one in Aeloria’s own early work. A marketing agency, a real case, anonymized here, was described in AI answers as an automotive repair shop. Not ranked lower. Not slightly out of date. A different business entirely, wearing the brand’s name. The engine’s error had nothing to do with the agency’s quality. It was about what the agency is, and every recommendation built on that foundation inherited the mistake.

In B2C and B2B you optimized to be found. In B2A you optimize to be understood. The entire discipline of GEO is, at its core, the retraining of mistrained buyers.

Then comes A2A, and the human disappears entirely

B2A still ends with a person acting on the AI’s recommendation. A2A removes even that.

In late 2025, OpenAI and Stripe launched the Agentic Commerce Protocol and Instant Checkout, letting Americans buy from Etsy sellers directly inside ChatGPT; Walmart put roughly 200,000 products into the chat window. Within six months, OpenAI had already reworked the model, because in-chat checkout converted markedly worse than merchants’ own sites: the transaction moved back to the merchant while discovery stayed inside the AI. Google answered in January 2026 with the Universal Commerce Protocol, co-developed with Shopify, Walmart, Target and Etsy, and by April its governing council included Amazon, Meta, Microsoft, Salesforce and even Stripe. Meanwhile, Visa’s Intelligent Commerce completed its first hundreds of agent-initiated transactions in December 2025 and predicts millions of agent-completed purchases by the 2026 holiday season, and Mastercard’s Agent Pay is processing live authenticated agent transactions. McKinsey projects agentic commerce could reach up to one trillion dollars in orchestrated US retail revenue by 2030, and three to five trillion globally.

Read that sequence again. The standards are still fighting each other; the direction is not. The payment networks, the most conservative institutions in commerce, are building rails for buyers who are not people.

And consumers are already further along than you’d think. Visa’s own research finds that nearly half of US shoppers, 47%, now use AI tools for at least one shopping task. Give it three years.

In an A2A transaction, your product page is read by a parser. Your pricing is compared by an algorithm. Your reputation is a set of signals scattered across the open web, weighed in milliseconds. There is no banner ad that reaches this buyer. A mistrained buyer in an A2A world doesn’t just misinform a customer. It completes the wrong transaction.

The question my dinner companion couldn’t answer

At the end of that dinner, I asked him: does he know what the AI engines say about his company right now, today, when a customer asks?

He didn’t. Neither do most. In Semrush’s 2026 AI Visibility Index, 45% of marketing leaders admitted they cannot accurately measure how they appear in AI-generated answers, and only 9% have the tools to track it across platforms. The rest can quote their Google ranking for twenty keywords and their cost per click to the cent, and have no idea whether the engines answering their customers were trained right.

If they’re wrong, incomplete, or you’re invisible, now you know what you’re dealing with. Not a missing tweet. A mistrained buyer, talking to your entire market, every single day.

The third letter changed. The businesses that notice first will own the answer. Everyone else will be a footnote in it, if they’re lucky.