Over the last year and a half, the reasoning models of the world’s leading AI labs have grown more sophisticated, better at adhering to instructions, and capable of completing multi-step processes. That has resulted in a surge in demand for AI agents, bots capable of autonomously executing 10-plus-step processes on behalf of users.
Agents are taking over the internet. Last year, automated traffic spiked 23.5% from the previous year, eight times the rate of human traffic growth, according to research from Human Security. Retail sites have seen especially high surges, with traffic from AI sources up almost 400% year-over-year in the first quarter of 2026, per Adobe.
Historically, agents weren’t able to connect to external servers or complete commerce transactions—the pipes simply weren’t there. To enable AI agents to connect to external tools and data sources without existing APIs or custom integrations, Anthropic in late 2024 established Model Context Protocol. MCP, overseen by the Linux Foundation, which has since become the de-facto technical standard for agent-to-tool compatibility; it’s the universal adapter that allows agents to move across the web and work across different environments.
In the year and a half since the debut of MCP, there’s been a proliferation of technical standards—many of which are enabled by MCP—designed to facilitate agentic advertising and commerce. Brands, agencies, platforms, and publishers have expressed growing interest in agentic advertising and commerce, which could theoretically enable simpler navigation of the fragmented programmatic ecosystem and lead to more efficient, transparent, and cost-effective transactions.
But the explosion of specs is a lot to keep up with. Here’s a simple guide to some of the leading technical standards and how they differ. Note that this list is by no means exhaustive and there is a longtail of ad and commerce standards emerging from various industry players.
The infrastructure layer
Model Context Protocol (MCP)
Rolled out in November 2024 by Anthropic and now governed by the Linux Foundation, MCP is the foundational standard that enables AI agents of all kinds to connect to external tech, data, and services. It works by linking an MCP client, which is typically an agent built on a major large language model, with local data sources or remote servers, allowing them to speak the same language.
Agent2Agent Protocol (A2A)
Introduced by Google in April of 2025, A2A is like MCP in that it is universally accessible and now managed by the Linux Foundation. However, rather than helping agents connect to external tech and data, A2A is a universal standard to agent-to-agent interactions. It allows agents to connect and communicate without exposing their internal logic, memory, or technical specs. A2A has gained widespread support from major SaaS players, including Microsoft, Amazon Web Services, SAP, Salesforce, and IBM. A2A was built atop a variety of popular standards including HTTP, SSE, and JSON-RPC.
Agentic advertising protocols
Ad Context Protocol (AdCP)
Created in October of 2025 by a consortium of organizations including Yahoo, PubMatic, and Optable—and helmed by Scope3 CEO and AppNexus co-founder Brian O’Kelley—AdCP is an open-source protocol that allows agents to communicate, as well as plan, negotiate, and execute media buys across different ad platforms. Built on MCP, AdCP is meant to standardize agentic digital ad transactions. The standard is overseen now by an independent industry group, AgenticAdvertising.org.
Agentic RTB Framework (ARTF)
Designed in November 2025 as an alternative to AdCP, the IAB Tech Lab-created ARTF embodies a different approach to agentic media buying. Whereas AdCP provides a shared language for agents to negotiate ad campaigns with platforms, ARTF seeks to update the infrastructure of real-time bidding to accommodate both agents and more efficient standard media buys. It does so by offering a containerized framework that lets agents operate within existing SSP and DSP environments but promises lower latency auctions.
“It is doing overdue modernization [of real-time bidding infrastructure], as well as opening the gateway for the agentic world to be able to transact with the supply chain,” Shailley Singh, executive vice president of product and chief operating officer at IAB Tech Lab, explained to ADWEEK upon the launch of the framework.
ARTF has since become the keystone of the IAB Tech Lab’s wider Agentic Advertising Management Protocols (AAMP), an initiative that houses a range of agentic protocols for connecting buy- and sell-side tech, agentic guardrails, and an agent registry system.
Agentic Audiences (formerly User Context Protocol)
Introduced originally by LiveRamp last year and donated to the IAB tech Lab, UCP, later rebranded as Agentic Audiences, provides a framework for agents to exchange user-level data in a privacy-safe way. Using the infrastructure, one agent can speak to another and swap identity and contextual signals—in essence, information about who a user is, what they are doing, and how they respond to given content—without exposing sensitive data on the user.
And instead of trading huge datasets about the user, Agentic Audiences lets agents use condensed machine-readable embeddings, which allows interpret data faster and with reduced energy expenditure. The framework is part of the IAB Tech Lab’s AAMP. It has been integrated with OpenRTB and Prebid to help seed industry adoption.
Agentic Mobile
Created by CloudX before being donated to the IAB Tech Lab and folded into AAMP, Agentic Mobile takes the logic of agentic digital ad transactions and translates it to mobile.
Other lesser-known standards, such as Agentic Ad Object (a derivation of AdCOM), and various buyer and seller reference agents, also live within AAMP alongside ARTF, Agentic Audiences, and Agentic Mobile. These are part of a management layer that establish rules for how buy- and sell-side agents discover, negotiate, and complete orders.
Agentic commerce protocols
Agentic Commerce Protocol (ACP)
Introduced in September of last year by OpenAI and fintech company Stripe, ACP launched in tandem with ChatGPT’s Instant Checkout feature, letting autonomous agents assist users in completing purchases within the chatbot’s interface. The open standard provides the connective tissue between AI, merchant, and payment systems, enabling payment detail verification, order confirmation, and purchase approval.
In March of this year, OpenAI announced that Instant Checkout purchases would migrate to third-party applications. Now, users can discover products inside the ChatGPT experience, but OpenAI will no longer own the entire transaction; purchases will be handled on the retailer’s own site or app. ACP still links ChatGPT to merchants.
Universal Commerce Protocol (UCP)
Debuted in January, UCP is designed to support agentic commerce across the shopping journey from product discovery and price negotiations to checkout and post-purchase support. It sets out a universal system for agents and tech to work together across consumer interfaces, business platforms, and payment tech. Created by Google in conjunction with a commerce partners iShopify, Etsy, Wayfair, Target, and Walmart, UCP is compatible with a variety of other agentic protocols, including MCP, A2A, and AP2.
Amazon, Meta, Microsoft, Salesforce, and Stripe all joined the UCP Tech Council in April 2026. Today, UCP is by and far the most dominant commerce protocol.
Agent Payments Protocol (AP2)
Created by Google in collaboration with Coinbase, PayPal, Mastercard, American Express, and others, AP2 lets AI agents securely exchange information and conduct commerce activity on users’ behalfs. It is built to keep real people in control, with mechanisms for verifying both the fact that a given user authorized a specific purchase (confirmed with a series of cryptographically signed mandates), and that the payment information provided is valid (confirmed with the help of the infrastructure partner that actually transfers funds). AP2 provides the answer to both of these factors in a format that any network or merchant can easily verify. AP2 is positioned as a payments layer add-on to A2A and MCP.
x404
Established by Coinbase and Cloudflare and folded under management of the Linux Foundation in April of this year, x402 is a reimagined, crypto- and AI-friendly version of the HTTP 402 “Payment Required” status code. The standard lets AI agents make payments using stablecoins. Rather than requiring a user or agent to sign up for an account, input their payment information, or manage an API, an agent or or human user can pay for a product or resource directly inside a standard HTTP request workflow using stablecoins, primarily USDC. It is blockchain-agnostic.
Founding members of the protocol include Google, Mastercard, AWS, American Express, Microsoft, Stripe, Circle, and Fiserv. Google integrated x402 into AP2 to act as the default stablecoin transaction rail last September.
Trusted Agent Protocol (TAP)
TAP, released in October 2025 by Visa in partnership with Cloudflare, creates a secure connection between agents and merchants throughout each step in an agentic commerce transaction. It is also meant to help merchants distinguish between trustworthy agents working on users’ behalfs and malicious or fraudulent bots. With TAP, an agent signs up and gets a cryptographic key tied to its unique identity; then, for each request that an agent sends using that key, the given merchant or infrastructure provider can check a reliable registry to pinpoint which agent sent the request, ensure its validity, and make sure that agent has been authorized to take actions. If all these criteria are met, the merchant can allow the agent to browse or check out. TAP relies primarily on traditional web cryptography, the most central being HTTP message signatures. Early adopters of include Coinbase, Microsoft, Shopify, Stripe and Worldpay.
Mastercard Agent Pay
Announced in April of 2025, Mastercard Agent Pay acts as a payments and trust protocol, similar to AP2, TAP, and x402. ith Agent Pay, Mastercard swapped a user’s card with a unique token assigned for an approved agentic interaction. Then, the system proves to the merchant that a real user authorized an action and that the agent followed the instructions it was given. Users remain in control, setting their own permissions and spending limits, helping to keep agents in line with authorized decisions only. The company is working with leading financial institutions like Citi, U.S. Bank, and other global organizations to make the tech more widely available, and in mid-2025, teamed up with Google to use UCP as the underlying common language for agentic commerce.
Agentic Merchant Protocol (AMP)
Introduced in March by agentic commerce platform Azoma, ACP is an enterprise-focused merchant protocol that offers a centralized system that brands and retailers can use to easily establish how their product catalogues are understood and acted upon by AI agents. The idea is that while platform-tailored standards like OpenAI’s ACP and Google’s UCP enable seamless agentic transactions, but don’t give brands control over how their products are represented or evaluated. It’s possible, in theory at least, that systems like ACP and UCP incorrectly score product data or even get product information wrong. AMP, which isn’t meant to replace those standards but work in collaboration with them, was created to act more as a brand governance layer, letting merchants establish a central source of truth about their products to then feed into other protocols.
Early adopters of AMP include Mars, Unilever, Reckitt, and L’Oréal.