AI assistants as commerce intermediaries

This is a bet that the winning consumer AI assistant will look less like software sold seat by seat, and more like a new checkout and discovery layer that sits between users and merchants. Once an agent is trusted to book flights, buy tickets, cancel subscriptions, or complete purchases across apps, it can take a cut of each transaction and sell merchants placement where buying intent is already explicit.1 2 3 4 5

• The early proof point is transaction volume, not subscriptions. Instinct is already reported as approaching $1B in annual transaction volume, with about half from travel, which shows that a personal agent can influence meaningful consumer spend before it has a mature paid software business.1 2
• The product mechanics are concrete. Perplexity partnered with PayPal and Venmo to let users pay for products, travel, and tickets inside chat, which turns an assistant from a research tool into a checkout surface. That is the template for how transaction fees can start attaching to agent usage.2 6
• Distribution determines who gets the commerce upside. Dots uses its own cloud computer and more than 4,000 plugins, while Meta positions Muse inside WhatsApp and Google is wiring Gemini into booking and local service apps like Ticketmaster, OpenTable, Angi, Thumbtack, and Zocdoc. The assistant that becomes the default front door for these tasks owns the merchant relationship.1 3 4 5

The next phase is a shift from chat monetization to agent monetization. As assistants handle more high intent workflows, revenue should track gross spend flowing through the agent and the ad budgets merchants devote to winning that demand, which will favor the platforms with the most trust, integrations, and daily habit.1 3 4 5 6