On November 11, 2026, the venture firm a16z published "Welcome to the B2A Economy," arguing that the internet's primary users are shifting from people to AI agents. Agents already account for most web traffic, write more than half of newly published articles, and produce most of the code pushed to production; and as agents move from answering questions to doing things on our behalf — deciding what to buy, what to subscribe to, and where to spend — how businesses win customers has to change with them.
The numbers first: agent traffic is already too big to block
a16z offered a set of figures: AI-driven traffic to Shopify stores grew 8x year over year in the first quarter of 2026, while orders from AI-powered searches grew nearly 13x; Human Security measured agentic browser traffic growing 8x between August and October, with Muse accounting for 60% and OpenAI's products for a combined 30%. Consumers referred by AI convert better and bounce less, and about 11% of shopping journeys in the U.S. passed through AI before reaching a transactional site. The article's conclusion: merchants used to build walls to keep bots out; now they have to compete for the bots' business.
From searching for you to buying for you: identity, payments, and connectors are the key
For now, most agents still work through the browser: taking over the user's browser or spinning up their own virtual machine and clicking around like a person. But a16z argues the real shift lies in direct access — connectors, plug-ins, and MCP — the equivalent of an app store for agents. The article gives an example: ChatGPT made a video ad using Krea, Descript, and ElevenLabs, coordinating the three products itself and even upgrading Descript to a paid plan to hit the deadline. OpenAI recently launched business features inside ChatGPT, letting companies claim an identity and upload product information; "Sign in with ChatGPT" is also on its way, aiming to let agents complete sign-ups and payments for users. Payments are catching up too, for example with Google's AP2 payments protocol.
Who cashes in first: four kinds of companies, and a ready-made example
a16z sorts the beneficiaries into four groups: the agents and models themselves — ChatGPT already has 1.2 billion weekly active users; new products made possible by agents, such as self-running small businesses; incumbents that optimize for agents, moving from search-exposure GEO toward helping agents evaluate and buy; and infrastructure for payments, fraud prevention, identity, and communications — volumes are still small, yet these links are already under strain. The ready-made example is coding agents: as of September 2026, more than 40% of GitHub pull requests were drafted by agents, up from 17% six months earlier, and more than 60% of new Supabase databases are created by AI tools. The article concedes, though, that most companies still block agent traffic over fraud and chargeback worries, and agent shoppers remain shut out at many online stores: agent shoppers blocked at the store door. a16z's view is that the door will open sooner or later.
