Akamai Technologies (NASDAQ: AKAM) and Visa have announced a strategic partnership to enhance identity, user recognition, and anti-fraud controls in what both companies describe as the next wave of digital commerce: agentic commerce, a scenario where AI agents browse, compare, and execute purchases on behalf of individuals.
The starting point is straightforward but challenging for any online store: as legitimate automated traffic (agents purchasing with permission) grows, so does the difficulty of distinguishing it from malicious automated traffic (fraud bots, scraping, promotion abuse, or attacks). Akamai and Visa’s proposal aims to address this blurry boundary: allowing merchants to verify who the agent is, what its intent is, and which real user it represents, without sacrificing control over security, personalization, or checkout experience.
What each contributes: Visa’s protocol + edge intelligence from Akamai
The partnership revolves around the Visa Trusted Agent Protocol, an authentication framework designed for AI agents to demonstrate they are authorized for a specific mission (e.g., “purchase X product with spending limit”), provide signals about the consumer they represent, and fit the payment into the merchant’s preferred flow (including options like network tokens or micropayments, depending on the case).
Akamai, in turn, integrates this protocol with capabilities already used in large-scale e-commerce environments: behavioral intelligence at the edge, user recognition, and protection against bots and abuse. The goal is for merchants to receive signals “before” interactions reach sensitive systems, enabling them to treat verified agents differently from bots trying to sneak through.
Patrick Sullivan, CTO of Security Strategy at Akamai, summarized it as the “dual identity problem”: proving who the agent is and, crucially, who the person behind it is. On Visa’s side, Jack Forestell (Chief Product & Strategy Officer) emphasized that agentic commerce can only scale if the ecosystem trusts the agents involved.
Why now: AI bots, retail pressure, and a new “type” of traffic
The backdrop is the rise in automated fraud and abuse. Akamai cites its 2025 Digital Fraud and Abuse Report, stating that AI-powered bot traffic has surged by 300% year-over-year, and the sector experienced over 25 billion AI bot requests over just two months. While these figures are aggregated, they highlight why merchants are wary of unchecked automation they can’t classify or control.
Meanwhile, Visa has been pushing for a standard protocol allowing agents to operate “with credentials” without turning the checkout into a security vulnerability or an ad hoc integration mess. Outlets like Axios frame this trend as an effort to “set rules” for how agents interact with payments and merchants, involving tech partners at the web’s perimeter.
The promise (and commercial interest): accepting agents without risking fraud
Akamai and Visa guarantee that the collaboration will enable merchants to:
- Better identify the agent and its intent (navigating is not the same as making a payment), combining signals from the protocol with anomaly detection and behavioral analysis.
- Link the agent to the underlying user while maintaining risk context, trust signals, and account information used for fraud prevention.
- Ensure more predictable payments in a world where the “buyer” isn’t always a human with a browser, but an agent operating under set limits and permissions.
The initiative also has a scale perspective: Visa asserts that the protocol design aims to minimize infrastructure and UX changes, facilitating adoption across its global acceptance network, which it estimates at 175 million merchant locations. Akamai adds significant weight: it states that 9 of the world’s top 10 retailers rely on its platform for performance and protection of digital commerce.
A market in flux… but without a “definitive manual” yet
While the announcement aligns with the narrative of “the next era” of e-commerce, agentic commerce remains in the early stages: what minimal data must an agent share to be trustworthy? How are limits and consent audited? What happens if a verified agent behaves anomalously due to a model error or malicious prompt?
This is where the Akamai-Visa collaboration seeks to position itself: not as marketing fluff, but as a practical identity and anti-fraud layer at the perimeter, filtering and classifying traffic. If this layer functions effectively, merchants’ incentives are clear: accept “good” automation (more conversions, greater convenience) without incurring higher fraud, returns, or promotion abuse.
via: Akamai