UCP vs ACP: Which Protocol Will Shape Agentic Commerce in 2026?

A neutral comparison of UCP and ACP for retailers in DACH and the EU

Executive Summary

Agentic Commerce describes the development where AI agents make purchasing decisions, execute transactions, and manage service processes on behalf of consumers or businesses. Two competing protocols shape the discussion: Google's Universal Commerce Protocol (UCP) and OpenAI's Agentic Commerce Protocol (ACP).

Key statements of this article:

UCP pursues an open, interoperable approach with dynamic capability negotiation between agents and merchants. The protocol was announced in January 2026 together with Shopify, Etsy, Wayfair, Target, and Walmart, aiming for standardized infrastructure across the entire commerce lifecycle.

ACP was introduced in September 2025 by OpenAI and Stripe, focusing on rapid integration through standardized schemas. The protocol enables Instant Checkout in ChatGPT and prioritizes low barriers to entry.

For merchants in DACH and the EU with complex checkout requirements such as VAT logic, right of withdrawal, SCA authentication, and local payment methods, UCP potentially offers more flexibility. ACP enables faster market entry but reaches limits with European checkout complexity.

Broad adoption in DACH and the EU is not expected for 2026. The realistic perspective: pilot projects with early adopters, limited reach, high learning curve. This article provides an overview for orientation, not implementation advice.

Context: Agentic Commerce is Infrastructure, Not a Feature

Why Standards Matter More Than Individual Products

In the history of e-commerce, standards have often changed more than individual products or platforms. HTTP enabled the web, SSL/TLS enabled secure transactions, OAuth enabled identity federation. Each of these standards created interoperability that did not exist before.

Agentic Commerce may be at a similar turning point. If AI agents are to shop for consumers, they need a common language to communicate with merchants. Without standards, every agent would need to be individually integrated for every merchant. That does not scale.

The question is not whether standards for Agentic Commerce will emerge, but which standards will prevail and how they will shift power dynamics in the commerce ecosystem.

Definitions for This Article

Agent: An AI system that acts on behalf of a user or business. In commerce context: a system that researches products, compares, makes or prepares purchase decisions, and executes transactions. Examples include Google Gemini with shopping capabilities, ChatGPT with checkout integration, or specialized shopping assistants.

Merchant: The retailer offering products or services. In protocol context: the system that responds to agent requests, provides product information, calculates prices, checks availability, and processes transactions.

Checkout: The purchase completion process. In agentic context: the question of where and how the final purchase takes place. Within the agent interface, on the merchant website, or in a hybrid model.

Identity: Identification and authentication of buyers. In agentic context: how an agent proves a user's identity to the merchant, manages consent, and authorizes payment methods.

Post-Purchase: All processes after purchase. Shipment tracking, returns, customer service, complaints. In agentic context: how agents can manage these processes on behalf of the buyer.

Distinction: Conversational Commerce vs Agentic Commerce

Conversational Commerce describes shopping experiences via chat interfaces, messengers, or voice assistants. Humans remain in the decision-making role; technology supports.

Agentic Commerce goes further. The agent acts, decides, and transacts on behalf of humans. Humans define parameters and preferences; the agent executes. This fundamentally changes requirements for commerce infrastructure.

What Is Publicly Documented Today

UCP: Publicly Documented Facts

According to Google Developer Guides, the Universal Commerce Protocol was introduced in January 2026 as an open standard for communication between AI agents and commerce systems. The documentation describes the following core elements:

Business Profile: Each merchant provides a machine-readable profile describing their capabilities, supported product categories, payment methods, and services.

Capability Negotiation: Agent and merchant dynamically negotiate which functions are available in a session. This enables flexible integration without rigid schema requirements.

Core REST Endpoints: The protocol defines standardized endpoints for session management, product discovery, cart operations, and order processing.

Optional Embedded Checkout: For complex checkout flows, the merchant can provide their own checkout interface embedded in the agent experience.

According to Shopify Newsroom, Shopify is a co-developer of UCP. The announcement from January 11, 2026 names Etsy, Wayfair, Target, and Walmart as additional partners. Shopify describes UCP as the foundation to connect "any merchant to every AI conversation."

ACP: Publicly Documented Facts

According to the OpenAI Blog, the Agentic Commerce Protocol was announced on September 29, 2025, together with the "Instant Checkout in ChatGPT" feature. Stripe is named as the payment partner.

The OpenAI Commerce Documentation describes ACP as a standardized framework for product representation, cart management, and payment handoff. The protocol uses defined JSON schemas for product data, shipping options, and order confirmations.

According to OpenAI, Instant Checkout in ChatGPT supports a curated selection of merchants processed through Stripe. Checkout occurs within the ChatGPT interface without redirect to the merchant website.

Perplexity Instant Buy: Classification

According to Perplexity Help Center, Instant Buy enables purchasing products directly in the Perplexity interface. Checkout occurs within Perplexity, supporting Apple Pay and Google Pay. Publicly available information focuses on the US market.

Comparison Framework: UCP vs ACP Across 8 Dimensions

The following table compares UCP and ACP across central dimensions. Classification is based on publicly available documentation. Where documentation is missing, this is noted.

UCP: 6 Steps with Dynamic Negotiation

1. Agent retrieves merchant's Business Profile

2. Capability Negotiation: Agent and merchant negotiate available functions

3. Session is created with negotiated parameters

4. Cart operations within the session

5. Embedded Checkout handoff (for complex flows)

6. Order completion and confirmation

ACP: 4 Steps with Fixed Schemas

1. Agent retrieves product data in schema format

2. Cart and shipping options are queried

3. Payment handoff to Stripe

4. Order confirmation

The key difference: UCP negotiates capabilities dynamically per session and enables merchant-specific checkout flows. ACP uses predefined schemas for faster integration but less flexibility with complex requirements.

The Core Difference as Philosophy

Handshake Versus Form

A helpful metaphor for the difference between UCP and ACP: UCP works like a handshake. Agent and merchant get to know each other, exchange capabilities, and negotiate what is possible in this relationship. ACP works like a form. The merchant fills in defined fields; the agent processes them.

Both approaches have their justification. The question is which approach fits which use cases better.

Why Capability Negotiation Can Make Sense

Commerce is complex and heterogeneous. A fashion retailer with size guides, virtual try-on, and complex returns logic has different requirements than an FMCG retailer with standardized products and subscription models.

Capability negotiation allows mapping this heterogeneity without requiring a universal schema to cover all edge cases. The agent asks: "What can you do?" The merchant responds with their specific capabilities.

Where a Schema Approach Has Advantages

Schema-based approaches like ACP offer clear advantages for quick integration. If the defined schemas cover requirements, time-to-market is short. For merchants with standardized checkout flows and limited complexity, ACP may be the more pragmatic path.

The challenge arises when reality is more complex than the schema. Then flexibility to map edge cases is missing.

UCP Deep Dive: Flow, Sequences, Payload Examples

Sequential Flow

The typical UCP flow follows these steps:

Step 1: Discovery and Profile Fetch

The agent retrieves the merchant's Business Profile. This profile describes which capabilities the merchant supports.

Simplified example payload:

    {
      "merchant_id": "merchant-123",
      "name": "Fashion Store Berlin",
      "capabilities": [
        "product_search",
        "inventory_check",
        "cart_management",
        "embedded_checkout",
        "order_tracking",
        "returns_initiation"
      ],
      "supported_regions": ["DE", "AT", "CH"],
      "payment_methods": ["credit_card", "paypal", "klarna", "sofort"],
      "checkout_mode": "embedded"
    }

Step 2: Capability Negotiation

Agent and merchant agree on capabilities relevant for this session.

Simplified example:

    {
      "session_id": "session-456",
      "requested_capabilities": ["product_search", "cart_management", "embedded_checkout"],
      "negotiated_capabilities": ["product_search", "cart_management", "embedded_checkout"],
      "session_parameters": {
        "locale": "de-DE",
        "currency": "EUR",
        "user_context": "authenticated"
      }
    }

Step 3: Session Create, Update, Complete

Within the session, the agent performs operations: product search, cart management, checkout initiation.

Simplified example for Cart Update:

    {
      "session_id": "session-456",
      "operation": "cart_update",
      "items": [
        {
          "product_id": "sku-789",
          "variant_id": "size-m-black",
          "quantity": 1
        }
      ],
      "response": {
        "cart_total": 89.95,
        "currency": "EUR",
        "shipping_options": [
          {"method": "standard", "price": 4.95, "days": "3-5"},
          {"method": "express", "price": 9.95, "days": "1-2"}
        ]
      }
    }

Step 4: Optional Embedded Checkout

For complex checkout flows, the agent hands off to the merchant's Embedded Checkout. The customer sees a merchant-controlled interface within the agent experience.

This is particularly relevant for:

Bundles and Sets: When products are offered as packages with price advantages, bundle logic must be correctly reflected in checkout.

Variant Logic: Fashion retailers with complex size and color combinations, size guides, or virtual try-on need more than simple product selection.

B2B Tiered Pricing: Business customers with individual price lists and volume discounts require logic beyond standard schemas.

Compliance Requirements: Products with age verification or similar requirements can perform necessary checks in embedded checkout.

ACP Deep Dive: Flow, Sequences, Payload Examples

Sequential Flow

The ACP flow is leaner and follows defined schemas:

Step 1: Product and Offer Representation

The merchant provides product data in a standardized format.

Simplified example:

    {
      "product_id": "prod-123",
      "title": "Classic Cotton T-Shirt",
      "description": "Premium quality cotton t-shirt",
      "price": {
        "amount": 29.95,
        "currency": "EUR"
      },
      "variants": [
        {"id": "var-1", "size": "S", "color": "white", "available": true},
        {"id": "var-2", "size": "M", "color": "white", "available": true}
      ],
      "images": ["https://merchant.com/images/tshirt-1.jpg"]
    }

Step 2: Cart and Shipping Options

The agent creates a cart and retrieves shipping options.

Simplified example:

    {
      "cart_id": "cart-789",
      "items": [
        {"product_id": "prod-123", "variant_id": "var-2", "quantity": 1}
      ],
      "shipping_address": {
        "country": "DE",
        "postal_code": "10115"
      },
      "shipping_options": [
        {"id": "ship-1", "name": "Standard", "price": 4.95, "estimated_days": 5}
      ]
    }

Step 3: Payment Handoff and Order Confirmation

Checkout occurs through Stripe within the agent interface.

Simplified example:

    {
      "order_id": "order-456",
      "status": "confirmed",
      "total": {
        "subtotal": 29.95,
        "shipping": 4.95,
        "tax": 6.63,
        "total": 41.53
      },
      "payment_status": "completed",
      "confirmation_url": "https://merchant.com/orders/456"
    }

Advantages of the ACP Approach

The ACP approach offers clear advantages for quick entry. Merchants with Stripe integration can go live relatively quickly. The schema structure reduces implementation complexity.

Limits of the ACP Approach

Limits appear with merchant-specific checkout logic. EU-specific requirements such as withdrawal notice, differentiated VAT display, or local payment methods beyond Stripe are not detailed in current documentation.

Edge cases like partial deliveries, backorder handling, or complex promotions require workarounds or are not supported.

Reality Check 2026: Adoption and Reach

What Is Communicated

UCP: Partnership with Shopify, Etsy, Wayfair, Target, Walmart. Shopify integration announced as first implementation. Timeline for broad availability not specified.

ACP: Live in ChatGPT with curated merchant selection. Stripe as payment partner. Focus so far documented on US market.

Hypotheses for DACH and EU

The following statements are hypotheses based on analysis of protocol documentation and known EU commerce requirements. They are not predictions.

Hypothesis 1: EU checkout complexity will be a challenge for both protocols. VAT logic with different rates, right of withdrawal with defined information requirements, Strong Customer Authentication, and local payment methods like SOFORT, Giropay, or iDEAL require more than standard schemas.

Hypothesis 2: Identity and consent flows will be harder to solve in the EU. GDPR-compliant consent management, cookie consent in agent context, and data sharing with third parties are not trivial to map.

Hypothesis 3: UCP offers more room for EU compliance through Embedded Checkout but requires higher implementation effort.

Hypothesis 4: ACP enables faster start, but merchants will need to make compromises on checkout customization.

None of these hypotheses should be understood as legal assessment.

Merchant Readiness Check: Prerequisites to Benefit

Maturity Level 0: Fundamentals Missing

Characteristic: No documented APIs. Commerce data fragmented. Product information not structured. No programmatic interface for inventory or pricing.

Benefit Potential: None. Agentic Commerce is not implementable.

Effort: Fundamental commerce infrastructure work required.

Risks: Resources bound for fundamentals instead of innovation.

Typical for: Smaller merchants with legacy systems. No vertical focus.

Maturity Level 1: Commerce APIs and Data Model Present

Characteristic: Documented APIs for products, inventory, orders. Structured product data. Basic integrations with ERP or WaWi.

Benefit Potential: Limited. Read access for product information possible. Transactional flows require additional work.

Effort: Weeks to a few months for pilot integration.

Risks: APIs not designed for real-time requirements. Performance issues under load.

Typical for: Mid-sized fashion and FMCG merchants with modern infrastructure.

Maturity Level 2: Checkout Logic Modular

Characteristic: Checkout as separate service. Promotions, loyalty, and inventory connected via APIs. Shipping and payment logic modular.

Benefit Potential: Substantial. UCP Embedded Checkout integrable. ACP usable for standard flows.

Effort: Months for production-ready integration.

Risks: Edge cases in checkout logic not covered. Monitoring and alerting required.

Typical for: Larger fashion retailers, beauty brands with D2C focus, progressive FMCG companies.

Maturity Level 3: Real-Time Capability Complete

Characteristic: All commerce processes via APIs. Real-time inventory. Fraud detection integrated. Identity and consent programmatically controllable. Post-purchase flows automated.

Benefit Potential: High. Complete agentic commerce flows implementable.

Effort: Weeks to months for integration. Ongoing optimization.

Risks: High dependency on system stability. Complex monitoring required.

Typical for: Enterprise fashion retailers, beauty corporations, FMCG market leaders with own tech organization.

Implementation Effort: Honest, with Ranges

Realistic Time Horizons

Pilot: 4 to 8 weeks at Maturity Level 2 or higher. Limited product selection, controlled user group, learning focus.

Limited Rollout: 2 to 4 months. Extended product range, internal reviews, performance optimization.

Production Hardened: 4 to 8 months. Complete integration, monitoring, alerting, runbooks, team training.

These estimates are hypotheses based on comparable integration projects. Actual efforts vary significantly.

Dependencies

Platform: Shopify merchants will likely be able to use UCP fastest. Shopify as co-developer has an advantage in native integration.

Headless vs Monolith: Headless architectures offer more flexibility but also require more integration work. Monolithic platforms depend on vendor roadmaps.

PIM, OMS, ERP: Every integration is only as strong as the weakest interface. If product data in PIM is not structured, no agent can use it meaningfully.

Payment and Fraud Stack: Existing payment integrations must be compatible. Fraud detection must be able to evaluate agent transactions.

Clear Expectation

Agentic Commerce is not plug-and-play. Integration requires technical resources, process adjustments, and organizational readiness. Several weeks to months is realistic, not days.

Platform Comparison: Who Is Furthest in 2026

Shopify

As co-developer of UCP, Shopify will likely be earliest with native integration. The announcement indicates an integrated solution merchants can activate. Details on availability in European markets are not yet documented.

Shopware

As leading European platform, UCP integration would be strategically sensible. No public announcements on UCP or ACP. Shopware 6's API-first architecture would fundamentally enable integration.

Adobe Commerce (Magento)

No public announcements on UCP or ACP. The flexible architecture fundamentally enables custom integrations. Enterprise customers would likely need agency support.

commercetools

As headless platform, commercetools is architecturally well-suited for Agentic Commerce. No public announcements. The API structure would enable UCP integration.

SAP Commerce

No public announcements. Integration would likely require significant customization effort. Enterprise customers with SAP focus depend on SAP roadmap.

Custom Headless

Complete control but also complete responsibility. Custom implementations can experiment early with UCP or ACP but bear all development and maintenance effort.

Risks and Typical Negative Examples

Fragmentation: Maintaining Two Protocols

Merchants may be forced to support both UCP and ACP to reach all relevant agents. This potentially doubles development and maintenance effort. Consolidation to one standard is not foreseeable.

Measurability and Attribution

When purchases are initiated through agents, attribution becomes more complex. Which touchpoint led to purchase? Which marketing measures were effective? Existing analytics frameworks need adaptation.

Margin and Promotion Errors

Complex promotion logic must be correctly communicated to agents. Errors in API implementation can lead to wrong prices or invalid promotions. Margin losses are possible.

Returns and Customer Service Costs

When agents initiate purchases that customers did not enter with sufficient information, return rates may increase. Customer service must be able to trace agent-initiated purchases.

Brand Control Versus Platform Control

When checkout takes place in an agent interface, the merchant loses control over brand experience. The balance between reach and brand experience must be consciously decided.

Amazon Vacuum

Amazon pursues its own approach with Rufus for AI shopping within the Amazon ecosystem. Whether and how Amazon participates in open standards like UCP or ACP is not documented. A closed ecosystem logic could increase fragmentation.

Decision Framework for Decision-Makers

Central Questions Before a Decision

Which customer journeys are relevant? Not all journeys suit agent commerce. Routine repeat purchases have higher potential than emotional first purchases with high consulting needs.

What complexity exists in checkout? The more complex checkout logic, the more important UCP flexibility becomes. For standard checkouts, ACP may be sufficient.

What differentiation through loyalty, bundles, B2B pricing? If loyalty programs, bundle logic, or B2B price lists are central, capability negotiation becomes more important.

What risks through identity, consent, fraud? EU-specific requirements for data protection and payment security require careful analysis.

What resources for maintenance? Agentic Commerce is not a one-time project. Ongoing maintenance, updates, and optimization are required.

Three Action Options

Act Now: For Shopify merchants with low checkout complexity and available tech resources. Start pilot, learn, iterate. Investment: Dedicated team for 2 to 3 months.

Prepare: For merchants at Maturity Level 1 to 2. Consolidate commerce APIs, modularize checkout, improve data quality. No protocol decision yet, but increase readiness. Investment: Infrastructure work over 3 to 6 months.

Watch: For merchants at Maturity Level 0 or limited tech resources. Observe development, learn from early adopters, prioritize foundation work. No investment in Agentic Commerce before 2027.

Outlook H1 2026

Realistic Expectations

Broad adoption of Agentic Commerce in DACH and the EU is not expected for H1 2026. Realistic developments:

Shopify will roll out UCP integration for merchants. Adoption rate will depend on activation simplicity.

ChatGPT Instant Checkout will become available in European markets, with documented limitations on checkout complexity.

Pilot projects with early adopters will generate initial learnings. These learnings will be decisive for broader adoption.

Signals for Broader Adoption

More platform integrations: When Shopware, Adobe Commerce, or commercetools announce native UCP support.

EU-specific documentation: When UCP or ACP provide detailed guides for EU compliance.

Merchant success stories: When first merchants report measurable results from Agentic Commerce pilots.

Signals for Fragmentation

Proprietary protocols: When additional platforms introduce their own standards.

Slow convergence: When UCP and ACP do not move toward each other.

Regulatory hurdles: When EU authorities express concerns about identity or consent.

Glossary

ACP (Agentic Commerce Protocol) A protocol introduced by OpenAI and Stripe for standardised commerce interactions within AI interfaces, focusing on rapid integration through predefined schemas.

API-first An architectural approach in which all functions of a system are primarily exposed through programmatic interfaces.

Agentic Commerce Commerce scenarios in which AI agents research, compare, and initiate transactions on behalf of consumers or businesses, with degrees of autonomy ranging from assisted selection to independent purchasing decisions.

Capability Negotiation A dynamic negotiation process between an AI agent and a merchant system in which available functions are determined per session.

Embedded Checkout A merchant's checkout interface embedded within the experience of an external interface to accommodate complex checkout requirements.

Headless Commerce The decoupling of frontend and backend of a commerce system, connected via APIs, enabling independent development of both layers.

In-Chat Checkout Purchase completion within a conversational AI interface without redirection to a conventional shop website.

SCA (Strong Customer Authentication) Two-factor authentication for electronic payments mandated by the EU Payment Services Directive PSD2 within the European Economic Area.

UCP (Universal Commerce Protocol) An open standard initiated by Google for communication between AI agents and commerce systems, featuring dynamic capability negotiation between agents and merchants.

Sources

The following public sources were used for this article:

Google Developer Guides on UCP: Technical documentation of the Universal Commerce Protocol, published January 2026.

Shopify Engineering Blog: Post on UCP co-development and Shopify integration, published January 2026.

Shopify Newsroom: Announcement "The agentic commerce platform: Shopify connects any merchant to every AI conversation", 11 January 2026.

OpenAI Commerce Documentation: Documentation of the Agentic Commerce Protocol, published September 2025 and continuously updated.

OpenAI Blog: Announcement "Buy it in ChatGPT: Instant Checkout and the Agentic Commerce Protocol", 29 September 2025.

Stripe Newsroom: Announcement of ACP partnership with OpenAI, September 2025.

Perplexity Help Center: Documentation on Instant Buy, continuously updated.

This article was published on 13 January 2026. Information is based on publicly available sources at time of publication. Agentic Commerce is a rapidly evolving field. Current developments may differ from information presented here.

For more information on AI applications in e-commerce, visit our AI in E-Commerce page.

For a closer look at why the headless vs. integrated platform decision is far more than a technical question, see our guest post on headless in E-Commerce.

For a neutral, vendor-independent view of leading platforms, see our System Deep Dives series. For structured support in platform selection, visit our Platform Selection page.

How Shopify has translated these protocols into a concrete product is shown in our piece on the Shopify Agentic Plan. A sober reading for operational commerce can be found in our Agentic Commerce reality check. And what practical consequences merchants should derive from these protocols today is set out in our piece on the strategic consequences for commerce leaders.