Agentic Commerce Is Coming: What ACP and MCP Mean for Revenue
What is agentic commerce?
Agentic commerce is when an AI agent doesn't just recommend a product but completes the transaction — comparing options, evaluating fit, filling in forms, and paying — on a person's behalf. The human sets the goal ("find and subscribe me to the best invoicing tool for a freelancer"), and the agent does the shopping: it shortlists, decides, and acts. This is a meaningful step beyond AI search. AI search changed how people discover products; agentic commerce changes who executes the purchase. When the buyer is software, the criteria shift. An agent doesn't fall for a beautiful hero image or a clever tagline. It rewards clarity, structure, verifiable claims, and the ability to actually complete a task without a human in the loop.
What are ACP and MCP?
They're the emerging standards that make agent actions possible — the plumbing beneath the shift. MCP (Model Context Protocol) is an open standard for connecting AI models to external tools, data, and systems through a common interface. Instead of every model inventing its own way to talk to every service, MCP gives them a shared language. It's part of what lets an assistant move from talking about a task to performing it against real systems. Agentic Commerce Protocol (ACP) efforts aim to standardize the commerce layer specifically: how agents discover merchants, evaluate offers, and carry out transactions in a consistent, trustworthy way. Together with MCP, these standards are the early scaffolding of a machine-to-machine buying layer. The exact standards and their names will keep evolving, and no one should bet the company on a single protocol winning. The durable point is the direction: buying is becoming something software does, and it needs shared rails to do it.
Why should a founder care now, before it's mainstream?
Because readiness compounds and can't be bought overnight. The things that make you agent-ready — structured data, machine-readable pricing, a corroborated presence across the web, transactable flows, a clear and consistent description of what you do — all take time to build and, especially, to earn. Reputation across the sources models read is not something you switch on the week it starts mattering. There's also a defaults dynamic worth taking seriously. When an agent repeatedly picks a tool for a category, that tool becomes the default answer — and defaults are extraordinarily sticky, because the agent has no reason to re-litigate a choice that works. The companies that are legible and transactable early are the ones that get chosen while the behavior is forming, and being the incumbent default is a position competitors struggle to dislodge later.
Isn't agentic commerce still years away?
Full, mainstream, buy-anything-through-an-agent commerce may well be years out. But the underlying shift — from "humans click" to "agents act" — is already visible in AI search results, early agent checkouts, and the rapid spread of MCP integrations. The transition won't arrive on a single announced day; it will phase in category by category. That's precisely why acting now is cheap insurance rather than a gamble. The cost of getting agent-ready today is modest and mostly one-time. The cost of discovering you're invisible and unbuyable to machines the quarter it starts driving real revenue is a scramble you run while competitors are already the default. This is a classic 2–5 year bet: small, early, and asymmetric.
How do I know if I'm ready for agentic commerce?
The concrete proxy is transactability: could an agent, acting on a recommendation, actually complete a purchase on your site right now? That means machine-readable pricing, explicit plans, structured data describing your offer, a signup that doesn't require a human, and crawler access to the pages that matter. An Agentic Score measures exactly this alongside your visibility, so "are we ready?" stops being a vibe and becomes a number you can improve.
Frequently Asked Questions
- Is agentic commerce the same as AI search?
- No. AI search is about discovery — the model surfacing or describing products. Agentic commerce goes further: the agent completes the transaction. You can be visible in AI search yet fail at agentic commerce if an agent can't actually buy from you.
- What is MCP in simple terms?
- The Model Context Protocol is a shared standard that lets AI models connect to external tools and data through one common interface, so an assistant can act on real systems instead of only talking about them.
- Do I need to implement a protocol to be ready?
- Not initially. The highest-leverage first step is the fundamentals — machine-readable pricing, structured data, clear positioning, transactable flows — which make you ready regardless of which protocol wins.
- Will agentic commerce replace human buying?
- Not wholesale, and not soon. It will handle more routine discovery-and-purchase work, especially for software and repeat buys, while humans stay in the loop for higher-stakes decisions.
- Why act now if it's a multi-year trend?
- Because agent-readiness and reputation compound slowly and defaults are sticky. Starting early is cheap and positions you as the default answer while the behavior forms.