Three years after ChatGPT’s debut on November 30, 2022, AI has evolved from a “smart assistant” to something far more powerful: the invisible infrastructure of global commerce.
In the span of a few weeks, OpenAI, Google, and Perplexity all switched on real shopping inside their AI interfaces. At the same time, Visa, Mastercard, Stripe, and PayPal lit up new tokenized payment rails built specifically for agents, not browsers.
Search → click → cart → checkout is being replaced by a new pattern:
Describe intent → AI researches → filters → negotiates → pays.
The interface for commerce is no longer the webpage.
It is the AI agent.
Executives cannot treat this as hype. Gartner expects AI agents to intermediate 15 trillion dollars in B2B spending by 2028. The broader agentic AI market is forecast to grow from roughly 4.8 billion dollars in 2024 to 93.2 billion dollars by 2032, a 44.6% CAGR.
This is not an “experiment.” It is a rewiring of how demand finds supply.
The November 2025 flip: shopping moves inside the model
OpenAI’s new Shopping Research turns ChatGPT into a buying advisor that stays with the user from “I don’t know what to get” to “this is the exact product, size, and merchant to buy from.”
Key characteristics:
Runs on a specialized GPT-5 mini tuned for shopping scenarios
Asks clarifying questions on budget, brand preferences, and use case
Produces structured, skimmable buyer’s guides instead of generic lists
Routes to instant checkout via agent-ready payment rails
It looks like a small UX tweak. It is not. It is OpenAI formally stepping into the “demand origination + routing + payment” layer of commerce.
Google: 50 billion products, real-time inventory, and “Buy for me”
Google’s move is even more explicit.
The Shopping Graph now covers 50+ billion product listings, from big-box to local mom-and-pop stores, with about 2 billion listings refreshed every hour.
AI Mode in Search turns vague, 20+ word questions (“I need a warm, packable black winter coat under $350 that works for travel and Toronto winters