Tony Wood examines how returns become trust recovery events, subscriptions become delegation maintenance, and loyalty becomes reliability scoring when the customer is an autonomous agent..
| Dimension | Traditional UX | Agentic Experience Design (AXD) |
|---|---|---|
| Primary material | Attention and affordance | Trust and delegation |
| User state | Present, navigating | Absent, delegating |
| Design output | Screens and interfaces | Outcomes and constraints |
| Temporal model | Session-based | Relationship-based |
| Success metric | Task completion | Trust calibration |
In agentic commerce, returns are not customer service events - they are trust recovery events. When an agent-initiated purchase results in a return, the return process tests the merchant's trust recovery architecture. The agent evaluates: How quickly did the merchant acknowledge the issue? Was the return policy machine-readable and unambiguous? Was the refund processed without human intervention? Each return is a data point in the agent's trust model of the merchant. Merchants with fast, transpa
Subscription management becomes a delegation maintenance problem. The human principal delegates ongoing purchasing authority to an agent - 'maintain my coffee supply' or 'keep my software licences current.' The agent must continuously evaluate whether the subscription still serves the principal's interests: Has the price changed? Has a better alternative emerged? Has the principal's usage pattern shifted? Subscription renewal is not automatic - it is a trust decision the agent makes on each cycl
Traditional loyalty - emotional attachment, brand affinity, habitual purchasing - does not transfer to agents. Agents have no emotions, no habits, and no brand preferences. But a structural form of loyalty emerges: reliability-based return behaviour. An agent that has successfully transacted with a merchant multiple times, with accurate data, reliable fulfilment, and transparent policies, develops a trust history that reduces evaluation cost on subsequent transactions. This is not loyalty in the
Fulfilment is the moment where trust architecture meets physical reality. An agent can evaluate a merchant's data quality, pricing transparency, and return policies before purchase - but fulfilment quality can only be assessed after purchase. Delivery accuracy (correct item, correct quantity, correct condition), delivery timeliness (within the promised window), and delivery transparency (real-time tracking, proactive delay notifications) are the post-purchase trust signals that determine whether
The most consequential moments in commerce are not the moments of purchase. They are the moments after purchase. The return that tests whether the merchant's promises were real. The subscription renewal that tests whether the relationship still serves the customer. The fulfilment experience that tests whether the merchant's data was accurate. In human-browsed commerce, these post-purchase moments are managed through customer service teams, loyalty programmes, and retention marketing. In This essay examines the post-purchase problem in agentic commerce - the set of design challenges that emerge after an autonomous agent has completed a transaction on behalf of its human principal. Returns, subscriptions, loyalty, and fulfilment are not operational afterthoughts. They are the primary trust signals that determine whether an agent transacts with a merchant again. The post-purchase experience is where trust is tested, maintained, or destroyed. In traditional commerce, a return is a customer service event. The customer is unhappy. The merchant processes the return, issues a refund, and attempts to retain the customer through apology, discount, or improved service. The return is a failure to be managed, a cost to be minimised, a friction point in the customer relationship. Each return is a data point in the agent's evolving trust model of the merchant. A merchant that handles returns quickly, transparently, and programmatically demonstrates that its trust recovery architecture works. The agent's trust in the merchant may actually increase after a well-handled return - because the merchant has proven it can recover from failure. This is counterintuitive from a traditional commerce perspective, where returns are always negative. In agentic commerce, a well-handled return is a trust signal. Conversely, a poorly handled return is a permanent trust degradation. If the return process requires human intervention - the agent must escalate to its human principal, who must call a cu