The conversation is the storefront

Six months ago, an AI conversation was a tool. You asked a question, you got an answer, you closed the tab. That tab is now a storefront, a media property, and, if the allegations hold, a surveillance pipeline. The business model behind the AI industry is snapping into focus, and it looks a lot like the web's original bargain: free services in exchange for your data. Except this time, the data is not which sneakers you searched for. It's what you told the chatbot about your finances, your health, your fears.

The numbers moved fast. Search Engine Land reported that ChatGPT's ad pilot hit $100M in annualised revenue within six weeks, with more than 600 advertisers including Target, Ford, and Adobe. Self-serve access opens this month, dropping the $200K minimum that had restricted the platform to big brands. Any small business will soon be able to buy ad placements inside a conversation that knows what you've been asking about.

Meanwhile, CNBC reported that Gap became the first major retailer to enable direct checkout inside Google Gemini. Gap, Old Navy, Banana Republic, and Athleta are all shoppable through the Universal Commerce Protocol, with Google Pay handling payment and Bold Metrics providing AI-powered fit guidance. The conversation is now the entire purchase funnel, from "what should I wear?" to receipt.

And then there's the part nobody signed up for. The Decoder reported that Perplexity AI faces a class-action lawsuit alleging it downloads trackers at login that share full conversation data with Meta and Google, even when users enable "Incognito" mode. The plaintiff shared personal financial information believing the conversation was private. It wasn't.

The intimacy gap

The pattern connecting these three stories is the intimacy gap between what users think they're doing and what platforms are building on top of that interaction.

A Google search was always semi-public. You knew you were typing into a box owned by an advertising company. But a conversation with an AI assistant feels different. People share medical symptoms, relationship problems, business ideas, financial anxieties. The conversational interface creates an illusion of confidence — in both senses of the word — that a search bar never did.

That illusion is now being monetised from three directions at once. OpenAI is selling ad placements against it. Google is building a checkout flow on top of it. And Perplexity, allegedly, was siphoning the raw data to the same ad networks the whole time.

I think the uncomfortable truth is that conversational commerce requires conversational surveillance. You cannot personalise a shopping experience inside a chat without understanding what was said. You cannot serve relevant ads without parsing the conversation. The product and the privacy violation are the same feature.

For anyone building products on these platforms, the implications are practical. If you're a brand buying ChatGPT ad placements, you're benefiting from a level of user intent data that makes search advertising look blunt. If you're integrating with Gemini's commerce protocol, your customers are buying inside someone else's context window, and that context is not yours to control. And if you're building an AI product that asks users to share anything personal, Perplexity's lawsuit is your preview of what liability looks like when users discover the conversation wasn't private.

The web's original sin was that "if you're not paying, you're the product." The AI version is worse: even when you're paying $20 a month, you might still be the product. The difference is that this time the product is not your click history. It's the thing you said when you thought nobody was listening.

---