Before checkout
AffixIO verifies the action before the merchant treats it as an authorized user decision.
Merchant checkout
How merchants verify AI agents at checkout
A verified agent checkout is a checkout where the agent, user mandate and purchase context are checked before the order is accepted.
For marketplaces
Marketplaces can check agent permission before inventory is reserved, sellers are notified or a paid order is created.
Where to put the check in your checkout
Place the permission check on the server before the action it controls. Do not rely on an agent-supplied claim, a browser badge or a successful check for an earlier basket. The merchant must decide which issuers and evidence it accepts.
- Resolve the actor and authority. Authenticate the request and retrieve the current user mandate. A recognised bot identity is not a spending mandate.
- Build the final transaction intent. Include the merchant, basket, total including tax and delivery, currency and any recurring commitment.
- Evaluate the exact intent. Check scope, expiry, revocation and remaining allowance against that transaction, using the supported verification integration.
- Enforce the outcome. Proceed only when the required checks permit it. A denial stops the action; an unresolved result needs the application's review or retry process.
- Request payment authorisation. The payment provider still performs its own checks. A successful permission check is not a successful payment.
- Retain the evidence reference. Link the decision to the order and payment attempt so support can reconstruct what happened.
Test the cases that break a happy-path demo
- Basket changes after approval: check the revised intent before submitting payment.
- The same request arrives twice: use your checkout's idempotency and order-state controls to prevent duplicate charges.
- Two purchases race for one allowance: reserve or consume allowance atomically in the system that owns it. Two independent successful checks must not imply two spending allowances.
- Verification times out: do not turn a network failure into an allow decision.
- Payment fails after verification: retain separate verification and payment states. A proof does not mean an order was paid.
These are merchant implementation responsibilities, not a claim that installing an SDK supplies an entire checkout. Start with bounded user consent, review the AffixIO SDK, and use the Agentic Pay Kit documentation for the supported integration.
Where AffixIO fits
AffixIO sits before the action. It helps your application ask whether the user, agent or delegated workflow is allowed to continue, then returns a decision your product can enforce.
For subscriptions
SaaS teams can verify plan changes, renewals, paid add-ons and usage increases triggered by agents.
For commerce
Retailers can check product category, basket value, return risk and user consent before fulfillment.
For APIs
API platforms can verify paid tool calls before compute, data access or downstream cost is triggered.
For disputes
Signed outcomes make later review clearer than a screenshot of an agent chat.
For privacy
The source record does not need to become a new permanent merchant database.
Designed for buyers who need trust without data sprawl
Use AffixIO when a checkout, platform, marketplace, API, internal tool or AI agent needs a clear permission result, signed evidence and a cleaner privacy posture. The value is not another dashboard full of personal records. The value is a decision with proof behind it.
Questions this page answers
Answers about verification, payment authorisation and implementation.
What is verified agent checkout?
It is a checkout flow that checks whether an AI agent is allowed to complete a purchase for the user before the order is accepted.
Who needs verified agent checkout?
Merchants, marketplaces, SaaS platforms, API providers and financial workflows that expect AI agents to trigger paid actions.
What does AffixIO return?
A decision your application can enforce, plus evidence for audit and support review.
Put a proof step before the risky action.
Start with the SDK, map the decision you need and connect the result to the place your product already enforces access, checkout or policy.
Personal Agent Protocol context
PAP-style sessions still need proof before action.
When a personal agent reaches a website, MCP tool, OpenAPI route or checkout flow, AffixIO can check the exact action and return yes, no or review evidence before the business lets it run.
PACT consent context
Scopes still need proof before execution.
PACT can identify the personal agent and carry user-granted scopes. AffixIO can check the exact action and return yes, no or review evidence before the brand lets it run.