Skip to content
TrueSession

What is agentic commerce? A guide for merchants

Agentic commerce is when AI agents search, compare and buy on a person’s behalf. What it is, how agents shop, the protocols involved, the risks for merchants and banks, and how to prepare.

Published , about 9 minutes to read

Agentic commerce, defined

Agentic commerce is buying and selling where an AI agent does part of the work a person used to do themselves: searching, comparing, asking questions, filling a cart, booking, and increasingly paying. The person sets a goal, such as a refundable flight under a budget, and the agent pursues it across businesses.

It differs from a chatbot on your own site in one important way. The agent works for the customer, not for you, and it may arrive from any AI assistant, browser or app.

How big is it?

Forecasts vary with their definitions, but they point the same way. Morgan Stanley estimates agentic shoppers could represent $190 to $385 billion of US e-commerce spending by 2030 (Morgan Stanley Research, Dec 2025). Bain puts the US market at $300 to $500 billion, or 15% to 25% of e-commerce (Bain & Company, Dec 2025). McKinsey sees $3 to $5 trillion of global consumer commerce orchestrated by agents by 2030 (McKinsey via Digital Commerce 360, Oct 2025).

The traffic is already visible. Adobe measured a 693% rise in generative AI traffic to US retail sites over the 2025 holidays (Adobe, Jan 2026), and Akamai found that AI bots made up 47.9% of commerce traffic on its network by December 2025 (Akamai via GlobeNewswire, Jul 2026). See our research page for more.

How AI agents shop

Agents reach businesses in three broad ways, often combined:

  • Reading pages. The agent fetches or renders your public pages and extracts what it needs, with no special integration.
  • Driving a browser. The agent operates a real browser, sometimes inside the customer’s own logged-in session, clicking and typing as a person would.
  • Calling an interface. The agent talks to an endpoint designed for agents, using an open protocol, and gets structured answers and actions.

The first two are hard to govern: the agent guesses from pages built for people, and the business cannot tell what the customer actually asked for. The third gives both sides structure, which is why the industry is investing in it.

The protocols to know

What changes for merchants

Your customer may never see your site. The agent reads, compares and decides. Accurate, current information becomes a sales channel, and stale pages become a liability.

Blocking has a cost. AI-referred visits convert well: Adobe reported they converted 60% better than other traffic in July 2026 (Adobe via Digital Commerce 360, Aug 2026). Turning every agent away gives up those customers.

Authorization becomes the hard problem. Payment systems assume a human clicks “buy”, as Google’s AP2 announcement points out (Google Cloud, Sep 2025). Federal Reserve Governor Waller named authentication, liability and fraud as the barriers to scaling agentic commerce (Federal Reserve, Sep 2026). Consumers agree: in a Gartner survey, at most 11% were willing to let AI make purchase decisions (Gartner, May 2026).

The legal line is unsettled. Amazon won a preliminary injunction against Perplexity’s Comet shopping agent in March 2026 (CNBC, Mar 2026), and the Ninth Circuit vacated it in August 2026 (US Court of Appeals for the Ninth Circuit, Aug 2026). Until rules settle, merchants need their own evidence of who did what, and with whose approval.

How to prepare: a merchant checklist

  1. Find out which agents already reach you and what they try to do.
  2. Decide what stays public, what needs a verified agent and what needs the customer.
  3. Give agents an official, structured route with live prices and reviewed facts.
  4. Verify agent identity cryptographically instead of trusting user-agent strings.
  5. Enforce protected data at your server or CDN edge, not only in browser code.
  6. Require the customer’s exact approval for purchases and account changes.
  7. Keep payment credentials on a human-only page, away from every agent.
  8. Keep a durable record of each task, approval and update for disputes.

Where TrueSession fits

TrueSession is a trust layer for that official route. It verifies agents, applies your rules and requires the customer’s exact approval before anything consequential, with a console to manage it. Core flows run in our laboratory today, and we are preparing the first pilots. See how it works, or read how to stop agents from acting without approval.