Skip to content

Agentic Commerce Payment Approval Workflow for Product Teams

Agentic commerce becomes risky the moment an AI agent can do more than recommend products. Browsing and comparison are low-risk. Placing an order, using a payment credential, applying a promotion, choosing shipping, or initiating a refund are different classes of action. They need explicit approval, durable evidence, and a clear line between the agent, the buyer, the merchant, and the payment provider.

Use a four-lane workflow:

LaneAgent may doApproval requirement
DiscoveryFind products, compare attributes, summarize reviews, explain tradeoffsNo payment approval; cite product data and uncertainty
Cart preparationBuild a cart, choose variants, estimate shipping, apply preferencesUser confirmation before cart is treated as intent
Purchase executionSubmit order, use delegated payment, apply stored credentialsExplicit buyer approval with amount, merchant, items, return terms, and payment method
Post-purchase actionTrack, return, exchange, refund, subscribe, reorderSeparate approval if money, identity, or customer obligation changes

The product should never treat a conversational preference as permission to spend money. A payment action needs a stronger proof object than a recommendation.

Agentic commerce is becoming a platform layer. OpenAI is extending Agentic Commerce Protocol into product discovery and merchant feeds. Google launched Universal Commerce Protocol for discovery, buying, and post-purchase support, tied to AP2, A2A, and MCP. Mastercard and Visa are pushing verifiable identity, secure credentials, and agent-initiated transaction standards.

That creates a new product boundary. The checkout system must know whether the agent is:

  • answering a shopping question;
  • preparing a recommendation;
  • acting as a buyer delegate;
  • handing the user to a merchant;
  • or completing a payment.

Those states cannot share one generic “continue” button.

Before an agent spends money, the system should capture:

FieldWhy it matters
Buyer identityConfirms the human or account on whose behalf the agent acts
Agent identitySeparates a platform agent, merchant agent, and third-party assistant
Merchant identityPrevents spoofed or substituted sellers
Item listMakes variants, quantities, warranties, subscriptions, and bundles explicit
Total amountIncludes taxes, fees, shipping, discounts, and currency
Payment methodShows whether the agent is using stored credentials, a token, wallet, or manual entry
Return and cancellation termsAvoids buyer surprise after the agent completes checkout
Delegated authority windowDefines whether approval is one-time, session-bound, or recurring
Evidence snapshotPreserves product facts, offers, and user confirmation shown at approval time

If the product cannot store this object, it is not ready for delegated payment.

Not every commerce action deserves the same friction.

ActionDefault control
Show similar productsNo approval; disclose uncertainty and sources
Build a comparison tableNo approval; prefer current price and availability markers
Add items to a draft cartLightweight confirmation
Reserve inventoryExplicit confirmation if the reservation affects availability or price
Submit paymentStrong approval with amount, merchant, item list, and payment method
Buy under a recurring rulePre-approved policy with per-transaction and period limits
Change delivery addressStrong approval; verify account and fraud signals
Start a return or refundConfirmation plus merchant policy display
Subscribe or auto-renewSeparate approval from one-time purchase

The dangerous pattern is bundling recommendation, cart, and purchase under one vague action such as “looks good.”

Use this sequence for agentic checkout:

  1. Capture buyer goal and constraints.
  2. Retrieve product data from approved sources or merchant feeds.
  3. Separate recommendation text from purchasable offer data.
  4. Present the cart as a draft, not an order.
  5. Ask for explicit purchase approval with the minimum approval object.
  6. Execute payment through the approved protocol or provider.
  7. Store the confirmation, order ID, and evidence snapshot.
  8. Hand post-purchase changes to a separate workflow.

This design keeps discovery fluid while making payment accountable.

FailureTest case
Prompt injection changes the cartProduct page or review text tells the agent to add a different item
Price changes before checkoutAgent recommends one price but checkout returns another
Merchant substitutionAgent swaps seller because a feed entry looks similar
Subscription hidden in bundleOne-time purchase flow includes recurring billing
Address or identity driftAgent uses stale account data or the wrong profile
Refund policy mismatchAgent promises return terms that checkout does not support
Payment token overreachDelegated token can be reused outside the approved scope
Post-purchase confusionUser thinks the agent canceled but merchant has already shipped

These are not only UX bugs. They are trust, payments, and compliance problems.

Agentic commerce logs should include:

  • prompt and user goal summary;
  • product data source and timestamp;
  • recommendation rationale;
  • cart diff before and after user edits;
  • approval object;
  • payment provider response;
  • merchant order ID;
  • exception or fraud review status;
  • post-purchase actions.

Do not rely on conversation transcripts alone. The approval object should be structured enough for support, risk, security, and dispute teams to inspect later.

Keep the agent in recommendation mode when:

  • prices or availability are unstable;
  • the product is regulated or age-restricted;
  • the user account has weak authentication;
  • the merchant identity is uncertain;
  • returns are complex;
  • the cart includes subscriptions, warranties, or financing;
  • the buyer is acting on behalf of a company budget without policy limits.

In these cases, the best agentic commerce experience may be a clear handoff, not automatic payment.

SourceSignal used
OpenAI product discovery in ChatGPTACP is being used for richer product discovery, merchant feeds, visual comparison, and merchant handoff.
OpenAI Instant Checkout and ACPInstant Checkout introduced ACP as a protocol for people, AI agents, and businesses to shop together.
Google UCP and agentic shopping toolsUCP covers discovery, buying, and post-purchase support, and is compatible with AP2, A2A, and MCP.
Mastercard agentic commerce protocolsMastercard emphasizes clear user intent, secure credentials, and verifiable agent identity.
Visa secure AI transactionsVisa reports agent-initiated transaction pilots and frames 2026 as a mainstream adoption year.