B2B sells to businesses and B2C sells to consumers, but both assume a human is reading and deciding. B2A (Business-to-Agent) assumes the buyer, or the buyer's proxy, is an AI agent. That single change reorders discovery, interface and transaction.
The comparison
| Dimension | B2B | B2C | B2A |
|---|---|---|---|
| Customer | A business, via a human buyer | A human consumer | An AI agent acting for a person or organisation |
| Interface | Salespeople, portals, contracts | Website, app, storefront | APIs, structured data, agent protocols |
| Discovery | Relationships, RFPs, referrals | Search, ads, social, browsing | Machine-readable feeds, structured data, agent search |
| What wins | Trust, terms, account management | Brand, experience, conversion | Legibility, structured trust signals, protocol support |
| Transaction | Negotiated, invoiced | Checkout, card payment | Agent-initiated, tokenised (ACP, AP2) |
| Optimised for | Human relationships | Human attention | Machine reasoning |
B2A does not replace B2B or B2C
It is a new layer over both. A B2C retailer still sells to consumers, but increasingly those consumers delegate discovery and buying to an agent. A B2B supplier still serves businesses, but procurement is being automated. B2A is the interface change, not a new customer segment. Analysis
Why the distinction matters
The tactics that win human attention, persuasive copy, imagery, funnel design, do little for an agent that reads raw data. Preparing for B2A means investing where agents actually look: machine-readable product data, agent-accessible interfaces, and being discoverable and recommendable to AI systems.