Marketing
UCP, ACP, and AP2: What Agentic Commerce Protocols Mean for Your Online Store
By: Courtney Turrin

Autonomous AI software is shifting from summarizing product reviews to executing purchases on behalf of consumers. To complete these transactions without human intervention, autonomous programs require standardized interfaces to discover inventory, configure shopping carts, and settle payments. These interface standards are known as agentic commerce protocols.
According to McKinsey research projections based on forecasted adoption rates and market conditions, the global agentic commerce opportunity could reach $3 trillion to $5 trillion by 2030, as outlined in their analysis of the agentic commerce opportunity. Transaction volume is already expanding rapidly. As reported by Signifyd across its specific merchant network data in October 2025, orders from AI agent referrals in its network increased by 1,247% year-over-year between October 2024 and October 2025.
For ecommerce leaders, capturing this demand does not require rebuilding your entire technical infrastructure. Understanding how the emerging autonomous commerce standards divide responsibilities allows you to prepare your catalog and checkout systems without disrupting existing consumer experiences.
What Are Agentic Commerce Protocols and Why Do They Exist?
Agentic commerce protocols are structured communication standards that allow autonomous software agents to locate products, verify real-time inventory, negotiate checkout parameters, and execute payments directly with merchant systems. They replace the fragile practice of AI crawlers scraping front-end website code designed for human eyes.
When human shoppers buy online, they interpret visual cues, dismiss promotional popups, select variants from drop-down menus, and manually enter credit card details. When autonomous AI agents attempt to navigate those same visual interfaces using automated browser scripts, they frequently fail. Website security systems mistake them for malicious scrapers, dynamic page elements load out of sequence, and session states expire.
Agentic commerce protocols solve this friction by establishing machine-to-machine agreements. Instead of clicking through a graphical user interface, an AI shopping agent interacts with a standardized protocol endpoint. The merchant system returns precise product availability, calculated shipping rates, and tax requirements in a structured data payload. The agent then passes cryptographically verified payment credentials to complete the purchase in seconds.
How UCP, ACP, and AP2 Compare Across the Transaction Lifecycle
The emerging agentic commerce landscape is organizing into distinct layers rather than a single monolithic tool. Three specifications define the primary stages of an automated purchase: the Universal Commerce Protocol (UCP), the Agentic Commerce Protocol (ACP), and the Agent Payments Protocol (AP2).
Transaction lifecycle
- Discovery & verificationUCPUniversal Commerce Protocol
- Catalog Feeds
- Real-Time Stock
- Merchant Specifications
- Negotiation & checkoutACPAgentic Commerce Protocol
- Cart Assembly
- Fulfillment Terms
- Intent Exchange
- Authorization & settlementAP2Agent Payments Protocol
- Cryptographic Tokens
- Identity Verification
- FIDO Rail
Universal Commerce Protocol (UCP)
Google introduced the Universal Commerce Protocol (UCP) on January 11, 2026, co-developed with Shopify, Etsy, Wayfair, Target, and Walmart, as detailed by Google for Developers. UCP operates primarily at the discovery and verification layer.
UCP establishes an open standard for how search ecosystems and shopping agents ingest catalog data, stock levels, and store policies. By unifying feed schemas and query interfaces across major retail systems, UCP ensures that an AI agent querying a product database receives real-time SKU availability, variant configurations, and merchant-verified attributes. This eliminates the hallucinated product details and outdated pricing common in unstructured scraping.
Agentic Commerce Protocol (ACP)
OpenAI and Stripe introduced the Agentic Commerce Protocol (ACP) as an open standard on September 29, 2025, alongside Instant Checkout in ChatGPT. ACP functions at the cart assembly and checkout coordination layer.
This protocol governs how a conversational assistant presents purchase options to a consumer and passes the buyer’s purchase intent to a merchant. Through Businesses in ChatGPT, introduced in beta by OpenAI to connect brands with the 1.2 billion people who use ChatGPT each week, an agent can initiate checkout, apply discount codes, calculate shipping tiers, and present an order summary directly within the chat interface. ACP standardizes the API calls required to assemble that order on the retailer’s backend without sending the shopper to an external checkout page.
Agent Payments Protocol (AP2)
Google continues to publish updates to AP2. AP2 operates at the payment authorization and settlement layer.
While ACP coordinates the transaction details, AP2 secures the payment execution. AP2 uses cryptographic passkeys and tokenized credentials so that a consumer can authorize an AI agent to spend up to a defined financial threshold. When the agent completes the transaction, it transmits a cryptographically signed authorization token rather than raw credit card details. This architecture gives merchants verifiable proof that the user authorized the delegated agent to make the purchase, mitigating fraud disputes and chargeback risks.
Protocol comparison
- UCPUniversal Commerce Protocol
- Backed by
- Google, Shopify, Etsy, Wayfair, Target, Walmart
- Transaction stage
- Discovery, catalog ingestion, inventory verification
- Core mechanism
- Standardized product schemas and inventory endpoints
- Primary benefit
- Accurate product retrieval across AI surfaces
- ACPAgentic Commerce Protocol
- Backed by
- OpenAI, Stripe
- Transaction stage
- Cart building, order parameters, conversational checkout
- Core mechanism
- Agent-to-merchant checkout APIs
- Primary benefit
- Direct conversion inside conversational platforms
- AP2Agent Payments Protocol
- Backed by
- FIDO Alliance, Google
- Transaction stage
- Payment tokenization, identity proof, transaction settlement
- Core mechanism
- Cryptographic authorization tokens and passkeys
- Primary benefit
- Fraud-resistant machine-delegated payment settlement
Together, these protocols create an interoperable framework that covers each stage of the transaction lifecycle.
Why AI Shopping Agents Fail at Traditional Checkout
The rapid introduction of agentic commerce protocols addresses an acute operational problem in online retail: standard web storefronts actively block AI purchasing attempts.
In an evaluation of 2,400 test purchases across 160 ecommerce websites conducted from August to October 2026, First Page Sage found that AI shopping agents failed to complete the order in 38.6% of attempts.
Agentic commerce by the numbers
- $3T to $5TProjected global agentic commerce opportunity by 2030McKinsey projection
- 1,247%Year-over-year growth in orders from AI agent referralsSignifyd merchant network, October 2024 to October 2025
- 38.6%Share of AI shopping agent checkout attempts that failed to complete an orderFirst Page Sage, 2,400 test purchases across 160 ecommerce sites, August to October 2026. Bar shows 0 to 100%.
These automated checkouts fail because standard ecommerce architecture is tuned exclusively for human behavior:
- Bot-Mitigation Firewalls: Enterprise security layers interpret headless browser actions as credential stuffing or content scraping attacks, triggering CAPTCHAs that autonomous software cannot solve.
- Dynamic DOM Renders: Single-page applications and client-side JavaScript updates frequently shift checkout form field selectors during rendering, breaking an agent’s field-mapping scripts.
- Variable Address Validation: Shipping address auto-complete tools and dynamic tax estimation scripts frequently require multi-step verification prompts that halt automated checkout scripts.
- Session Token Expirations: Multi-tab or multi-step checkout funnels designed to track human cookies often invalidate sessions when inputs arrive via automated sequential post requests.
Without protocol-level integration, online merchants experience higher server load from scraping bots while losing out on completed transactions. Standardized protocols bypass these front-end obstacles by routing purchases through dedicated, authenticated transaction endpoints.
Which Agentic Commerce Protocols Does Your Online Store Actually Need?
Ecommerce and marketing directors evaluating these protocols do not need to deploy extensive custom software engineering across every standard. The adoption path depends on how your platform handles technical feeds and checkout settlement.
Build or enable
- Native UCP via Merchant Center
- Platform-level ACP and Stripe Connect
- Core focus: data hygiene and feed sync
- Expose UCP-compliant REST or GraphQL feeds
- Implement ACP cart-session endpoints
- Deploy AP2 cryptographic token verification
For merchants on enterprise platforms such as Shopify or modern commerce architectures using payment gateways like Stripe, protocol interoperability is largely managed at the vendor layer. Platform providers are natively integrating UCP catalog formats and ACP checkout hooks into their hosted ecosystems.
Your operational focus centers on three practical implementation stages:
Readiness path
- RequiredStage 1Catalog interoperabilityUCP alignmentSyndicate structured product data and follow standardized merchant attributes. Required for any retailer selling physical goods.
- RecommendedStage 2Delegated cart executionACP supportEnable conversational or express checkout through your payment providers. Fits high-volume, repeat-purchase and impulse categories.
- MonitorStage 3Cryptographic settlementAP2 readinessWatch your payment processor’s roadmap for delegated passkeys and agent-tokenized billing.
Stage 1: Catalog Interoperability (UCP Alignment)
Every online retailer selling physical goods needs discovery-layer alignment. If an AI agent cannot verify product specifications, shipping weight, and live warehouse inventory, it will exclude your brand from consideration. You achieve this alignment by syndicating clean, structured product data to central shopping graphs and ensuring your product feeds adhere to standardized merchant attributes.
Stage 2: Delegated Cart Execution (ACP Support)
Online retailers competing in high-volume, repeat-purchase, or impulse consumer categories benefit directly from ACP. Enabling instantaneous checkout inside chat environments removes friction for buyers comparing products across platforms. If your payment infrastructure supports conversational or express checkout extensions through providers like Stripe, turning on delegated agent checkout provides an immediate conversion advantage.
Stage 3: Cryptographic Settlement (AP2 Readiness)
AP2 represents the security foundation that payment networks will require as autonomous purchasing volume scales. Because AP2 is governed by the FIDO Alliance, mid-size merchants do not need to author custom cryptographic verification systems. Instead, monitor your payment processor’s roadmap to ensure your checkout gateway supports delegated passkeys and agent-tokenized billing when these features roll out to standard merchant accounts.
Implementation Checklist: Preparing Your Ecommerce Architecture for AI Agents
To capture autonomous AI purchasing demand without destabilizing existing operations, marketing and technical teams can execute a straightforward architectural checklist.
Implementation checklist
- Step 1Structure real-time inventory and pricing endpointsUpdate variants, inventory and prices without multi-hour caching delays. Audit feeds for GTIN, MPN, brand name, dimensions and material.
- Step 2Standardize machine-readable store policiesPublish tables for shipping costs, delivery speed, regional restrictions and return windows. Remove surcharges that appear only at final checkout.
- Step 3Configure platform-level agent checkout integrationsCheck native channel integrations such as ChatGPT, confirm tokenized express checkout and whitelist authenticated agent IPs.
- Step 4Establish dedicated agent order attributionRecord referral channels, protocol origins and agent identification headers. Monitor completion rates, abandonment points and API error codes.
1. Structure Real-Time Inventory and Pricing Endpoints
AI agents do not tolerate stale data. If an agent attempts to purchase a variant that is out of stock, the transaction aborts instantly.
- Ensure product variants, inventory quantities, and active prices update across your product feeds without multi-hour caching delays.
- Audit product feeds to verify that mandatory attributes (GTIN, MPN, brand name, dimensions, and material data) are fully populated.
2. Standardize Machine-Readable Store Policies
Autonomous purchasing agents evaluate merchant policies before selecting where to buy. Ambiguous policies cause agents to recommend a competitor with clearer terms.
- Publish standardized, machine-readable tables detailing shipping costs, delivery speed thresholds, regional restrictions, and return windows.
- Eliminate hidden surcharge logic that only reveals itself on the final checkout screen, as unexpected fee injections cause automated agent checkouts to fail.
3. Configure Platform-Level Agent Checkout Integrations
Rather than building custom API connectors from scratch, use your existing software ecosystem to support agentic commerce protocols.
- Verify whether your ecommerce platform has enabled native integrations for conversational commerce channels like ChatGPT.
- Check that your payment gateway settings support tokenized express checkout configurations.
- Whitelist authenticated commerce protocol agent IPs within your security firewalls to prevent false-positive bot blocks on legitimate transactions.
4. Establish Dedicated Agent Order Attribution
Tracking agentic commerce requires separating autonomous transactions from standard organic and direct traffic.
- Configure your analytics and order management system to record referral channels, protocol origins, and agent identification headers.
- Monitor agent order completion rates, cart abandonment points, and API error codes to resolve operational checkout bottlenecks early.
Preparing your online store for autonomous purchasing is an exercise in data precision and platform readiness. To understand how these frameworks connect to broader AI SEO strategies and decipher the shifting landscape of AEO, AIO, GEO, and AISEO, marketing teams must align their data schemas across every channel. To evaluate your catalog architecture and build a clear roadmap for autonomous commerce, invite your leadership team to Get Clarity with a discovery call with an Xponent21 strategist.
