“Attribution Without Chaos”

Agentic Commerce Protocol (ACP) and SharedPaymentToken: What Brands Need to Know

ACP defines how AI agents transact with merchants. SharedPaymentToken enables secure agent-initiated payments. Learn what both mean for brands preparing for agentic commerce in 2026.

What Is the Agentic Commerce Protocol (ACP)?

Agentic Commerce Protocol (ACP) Defined: A technical framework that enables AI agents to discover, negotiate, and transact with merchants and services on behalf of users — without requiring the user to interact with a website or app interface. ACP defines the standards for how agents authenticate, access product data, process payments, and confirm orders in a machine-to-machine commerce environment.

The rise of AI agents capable of taking action — booking travel, purchasing products, renewing subscriptions — has exposed a fundamental gap in the current web: it was built for humans to browse, not for agents to transact. Agentic Commerce Protocol is one of the emerging frameworks designed to close that gap.

Why ACP Matters Now

When a user tells an AI assistant “reorder my usual coffee from last month,” the agent needs to do several things that current infrastructure doesn’t support cleanly:

  • Identify the right merchant and product
  • Authenticate as an authorized agent for that user
  • Access real-time pricing and availability
  • Process payment without redirecting the user to a checkout flow
  • Receive and store a purchase confirmation

Each of these steps currently requires fragile screen-scraping, proprietary API integrations, or the user manually completing the final steps. ACP aims to standardize these interactions so any compliant agent can transact with any compliant merchant.

SharedPaymentToken: The Payment Layer of Agentic Commerce

One of the most-discussed technical components in the ACP ecosystem is the SharedPaymentToken — a secure, reusable token that allows an AI agent to initiate payments on a user’s behalf without exposing raw payment credentials to third-party merchants.

How SharedPaymentToken Works

The SharedPaymentToken architecture works roughly as follows:

  1. Token issuance. A user authorizes their AI assistant (or the platform running the agent) to hold a payment token scoped to specific merchants, amounts, or categories.
  2. Agent-initiated payment. When the agent completes a transaction, it presents the SharedPaymentToken to the merchant’s payment endpoint rather than card details.
  3. Merchant verification. The merchant validates the token against the issuing payment network (Stripe, Visa, or similar) and confirms authorization.
  4. Confirmation loop. Both the agent and the user receive structured confirmation that the transaction completed.

The key benefit: the user’s raw payment credentials never leave the token issuer. The merchant never sees card numbers. The agent can transact on behalf of the user within pre-set parameters without requiring explicit approval for each purchase — but with full auditability.

ACP vs. UCP: What’s the Difference?

ACP is often discussed alongside the Universal Commerce Protocol (UCP), which originated from a different set of stakeholders. The practical differences:

Dimension ACP UCP
Primary focus Agent-to-merchant transaction standards Universal product data + commerce interoperability
Payment approach SharedPaymentToken model Varied — integrates with existing payment rails
Agent authentication DID-based agent identity OAuth-based delegation
Adoption stage (2026) Early — pilot implementations Early — growing enterprise interest

In practice, brands planning for agentic commerce readiness in 2026 need to watch both protocols — and understand that neither is fully standardized yet. The smart move is to implement the content and data layer that makes your catalog agent-readable regardless of which protocol wins.

What ACP Means for Brands and Merchants

If ACP reaches broad adoption, the implications for brands are significant:

Product Data Must Be Machine-Readable

Agents don’t browse product pages — they query structured data endpoints. Brands that haven’t invested in structured product data (clean schemas, real-time inventory APIs, machine-readable pricing) will be invisible to agent-driven purchase flows.

The Checkout Flow Changes Fundamentally

ACP assumes the agent completes the transaction. The traditional checkout UX — cart, shipping, payment entry — becomes a backend process the user never sees. Merchants need to expose headless commerce APIs that agents can call directly.

Brand Visibility Shifts from SEO to Agent Citation

When an agent decides which merchant to transact with, it doesn’t rank pages — it evaluates trust signals, reputation data, and structured product information. The question shifts from “do I rank on page one?” to “does my brand appear in agent-generated consideration sets?” This is exactly the problem that AI-driven brand visibility measurement is designed to track.

The Content Layer: ACP’s Missing Prerequisite

One aspect of ACP that receives less attention than the payment mechanics is the content layer — the structured, machine-readable representation of your brand, products, and policies that agents query before initiating a transaction.

An agent deciding whether to purchase from Brand A or Brand B will evaluate:

  • Product specifications (structured data, schema markup)
  • Return and warranty policies (machine-readable, not buried in legal PDFs)
  • Reviews and trust signals (accessible via API, not just HTML)
  • Real-time availability and pricing

Understanding what a content layer in agentic architecture looks like — and building one — is a prerequisite for ACP readiness, regardless of which payment protocol eventually wins.

Current State of ACP in 2026

As of early 2026, ACP remains in active development with no single authoritative specification. Key developments to track:

  • Stripe’s agent payment work — Stripe has published early documentation on payment APIs designed for AI agent use cases, which aligns with SharedPaymentToken concepts
  • Anthropic and OpenAI tool use — Both companies have expanded function-calling capabilities that enable agents to interact with commerce APIs
  • Schema.org extensions — Working groups are extending structured data schemas to support agent-readable product and service descriptions
  • Enterprise pilots — Several large retailers and travel companies have disclosed early tests of agent-initiated purchase flows

The prudent stance for brands in 2026 is to build agent-readable infrastructure now — clean APIs, structured data, machine-readable policies — rather than waiting for protocol finalization.

Frequently Asked Questions

What is the Agentic Commerce Protocol (ACP)?

ACP is a technical framework that standardizes how AI agents discover products, authenticate with merchants, and initiate transactions on behalf of users. It defines the APIs, data formats, and payment flows that enable machine-to-machine commerce without requiring human interaction with a checkout interface.

What is a SharedPaymentToken?

A SharedPaymentToken is a secure, reusable payment credential that allows an AI agent to initiate purchases on behalf of a user without exposing raw card or bank details to merchants. The token is issued by a payment platform, scoped to specific merchants or spending limits, and validated at transaction time by the payment network.

Is ACP the same as UCP?

No. ACP (Agentic Commerce Protocol) and UCP (Universal Commerce Protocol) are separate but related frameworks. ACP focuses primarily on the agent-to-merchant transaction layer and SharedPaymentToken mechanics. UCP focuses more broadly on product data interoperability and commerce standards. As of 2026, both are in early adoption stages.

What do brands need to do to prepare for agentic commerce?

The most important steps are: publish structured product data using schema.org markup, expose real-time inventory and pricing via API, make return and warranty policies machine-readable, and build an agent-readable content layer that represents your brand accurately to AI systems. Payment integration with agent protocols can follow once standards mature.

When will ACP be widely adopted?

No firm timeline exists as of 2026. Most analysts expect early mainstream implementations in specific verticals (travel, subscriptions, repeat-purchase consumer goods) within two to three years, with broader adoption following. Brands that build agent-readable infrastructure now will be positioned to participate in early rollouts.

author avatar
Will Tygart
Will writes about search, content strategy, and the shifting ground beneath both. His work focuses on SEO, AEO (Answer Engine Optimization), and GEO (Generative Engine Optimization) — the disciplines that decide whether content gets found by people, surfaced in answer boxes, or cited by AI systems. He genuinely enjoys the writing part. Most of what shows up here started as a question worth chasing.
Share the Post:

Unlock the Power of
Topic-Based Marketing

Topic Intelligence is a cutting-edge, deep-learning AI system designed to revolutionize your marketing strategy. Unlike traditional LLM-based tools, our advanced platform delivers actionable insights by analyzing topics that matter most to your audience. This enables you to create impactful campaigns that resonate, drive engagement, and increase conversions.