The B2A economy is a term coined by the investment firm a16z on November 10, 2026. B2A stands for Business-to-Agent: the next customer a business serves is no longer a person, but an AI agent acting on someone's behalf. Agents have moved from merely answering questions to comparing options, placing orders, subscribing, and paying—and what businesses now compete for is not human attention, but an agent's choice.
How Far Has Agent-Led Spending Actually Come?
The signs are more concrete than the slogan. Shopify disclosed that in Q1 2026, AI-driven traffic to its stores grew 8x year over year, and orders from AI-powered searches grew nearly 13x. As of September 2026, more than 40% of GitHub pull requests were drafted by agents, up from 17% six months earlier. ChatGPT has reached 1.2 billion weekly active users. On November 9, OpenAI launched Businesses in ChatGPT, letting merchants claim their storefront, upload product information, and see how users interact with their products.
The process itself is not mysterious: an agent first reads structured information—price, stock, reviews, return policy—then places an order according to the budget and preferences its owner set, with the human confirming only at key moments. What is changing is that the person who used to browse and compare before buying is becoming a piece of software.
How B2A Differs From B2B and B2C
| Who buys | How the decision is made | What businesses should optimize | |
|---|---|---|---|
| B2C | Individual consumers | Ads, word of mouth, personal taste | Storytelling, branding, attention |
| B2B | People inside a company | Tenders, price comparisons, approval processes | Relationships, proposals, service |
| B2A | AI agents | Reading structured data and comparing parameters by rule | Machine-readable product information, clear prices, and APIs |
B2A is not here to replace B2B and B2C. It adds a layer of intermediation on top of them: where businesses used to persuade people, they must now first be found, understood, and selected by an agent.
Don't Overstate B2A Just Yet
First, many businesses are still blocking agent traffic over fraud and chargeback concerns; a16z itself admits companies will only open up as this traffic becomes more valuable. Second, payments, identity, and liability are unsettled: whose identity does an agent use to order, who is responsible when it buys the wrong thing, and who handles the refund—there is no common standard yet. Third, the data a16z cites is that about 11% of shoppers in the US consult AI before a transaction. The trend is real, but it is still a long way from mainstream.
For merchants, making product information machine-readable—clear prices, stock, delivery, and return terms—matters more right now than buying another round of ads. For ordinary people, before letting an agent order on your behalf, set a budget cap and a rule such as "ask me before spending more than this amount"—then talk about convenience. a16z's full analysis of the B2A economy and why four tests of multiple agents working for users all lost to a single coordinator can help you judge how far this shift is from everyday use.
