Launch13 Oct 2026

Forter launches an engine to verify AI shopping agents and run headless agent checkout

Forter's Agentic Commerce Engine separates legitimate AI shopping agents from bots and lets merchants set rules for agent-driven checkout.

What changed

Fraud prevention provider Forter announced its Agentic Commerce Engine on 13 October 2026. The product is aimed at merchants who are starting to see AI shopping assistants browse, compare and buy on behalf of consumers, and who need to tell those agents apart from scrapers and fraud bots. The engine has three parts. A fraud module classifies agent and scraper traffic in real time, separating known assistants from abusive automation; Forter says this is live for its existing customers. An orchestration layer supports headless checkout through agentic protocols, so a merchant can let an approved agent complete a purchase without screen scraping and attach conditions to each agent transaction. A dashboard shows which agents are transacting on a merchant's site. All of it runs through a single, no-code integration for Forter customers. Forter backs the launch with its own network data: it says agent-driven orders have grown more than fourfold since August, while automated fraud has roughly doubled over the past two months. In a controlled test, it found headless agent checkout ran about ten times faster than an agent scraping the storefront. The launch matters because agent traffic breaks many of the assumptions fraud models rely on, such as device signals and human browsing patterns. Merchants that block all automation risk turning away real buyers, while those that let everything through invite abuse. Tools that verify agents and apply per-agent rules are becoming part of the checkout stack, alongside network programmes from Visa, Mastercard and American Express.

What merchants must do

Measure how much of your traffic and orders already come from AI agents before deciding whether to block or allow them. If you use Forter, ask about enabling agent classification and test per-agent rules on low-risk products first. Check that your fraud rules do not automatically decline legitimate agent orders that lack normal human browsing signals.