01 - The Empty Trading Floor
There is a photograph that captures the future of commerce more precisely than any forecast. It shows a trading floor - vast, architecturally magnificent, built for hundreds of human traders - entirely empty. The terminals still glow. The transactions still execute. But the humans have gone. The machines remained. This is not a metaphor. This is the literal trajectory of commercial exchange in the age of agentic AI.
We are entering an era where autonomous agents - software systems that can perceive, reason, plan, and act on behalf of humans and organisations - are becoming participants in commercial relationships. Not tools that humans use to transact. Not interfaces that mediate between buyer and seller. Participants. Entities that negotiate, purchase, subscribe, cancel, complain, and switch providers - all without human involvement in the moment of transaction. This is agentic shopping at its most fundamental: the delegation of commercial activity to autonomous systems.
This is agentic commerce. And it demands an entirely new design discipline.
02 - What Is Agentic Commerce?
Defining the New Commercial Paradigm
Agentic commerce is what happens when autonomous AI agents become participants in markets - not as tools that assist human shoppers, but as economic actors that discover, evaluate, negotiate, and transact on their own. Three distinct transaction patterns define the landscape: machine-to-human commerce (an agent purchasing from a human-operated business), human-to-machine commerce (a human purchasing from an agent-operated service), and machine-to-machine commerce (agents on both sides negotiating and settling without any human in the loop).
This is not a speculative future. Gartner predicts that by 2028, 15% of day-to-day work decisions will be made autonomously through agentic AI, and that by 2030, 25% of enterprise software agent interactions will involve machine customers - autonomous agents that discover, evaluate, negotiate, and purchase goods and services on behalf of their human or organisational principals.
The implications are architectural. Every commercial system - from e-commerce platforms to banking services, from insurance products to subscription management - must now be designed for a world in which the entity on the other side of the transaction may not be human. The interface may be an API. The decision-maker may be an algorithm. The loyalty may be calculated, not felt. And the trust architecture that underpins the relationship must be engineered, not assumed.
Commerce is no longer a human activity mediated by technology. It is becoming a technological activity governed by human intent.
03 - The Machine Customer
A New Category of Economic Actor
The machine customer is not a chatbot. It is not a recommendation engine. It is not a price comparison tool that a human uses to make a decision. A machine customer is an autonomous agent that holds delegated authority to make purchasing decisions, execute transactions, manage subscriptions, and switch providers - all within parameters set by its human principal but without requiring human approval for each individual action. The machine customer is the entity that performs agentic shopping - discovering, evaluating, negotiating, and purchasing on behalf of absent humans.
Consider the trajectory in financial services. Today, a customer logs into their banking app, reviews their savings rate, compares alternatives, and manually switches to a better product. Tomorrow, their financial agent monitors rates continuously, evaluates switching costs, assesses the reputational risk of the new provider, calculates the net benefit after tax implications, and executes the switch - notifying the customer after the fact. The customer's role shifts from operator to governor. They set the delegation scope, define the outcome specification, and calibrate the trust boundaries. The agent does the rest.
This creates three categories of machine customer that designers and businesses must now account for:
Bound Agents
Agents that act within tightly defined parameters set by a single human principal. They can execute transactions but cannot deviate from specified constraints. The standing order is the ancestral form.
Adaptive Agents
Agents that learn from patterns, adjust their behaviour over time, and make increasingly autonomous decisions as trust accrues. They operate within an operational envelope that expands with demonstrated competence.
Autonomous Agents
Agents that operate with broad delegated authority, making complex multi-step decisions across providers and product categories. They represent the full expression of the machine customer - an economic actor in its own right.
04 - The Five Shifts
How Agentic Commerce Transforms Every Commercial Assumption
The emergence of agentic commerce does not merely add a new channel. It fundamentally restructures the assumptions that underpin commercial design. Five shifts define this transformation:
Shift 01
From Interface to Intent
Traditional commerce is designed around interfaces - websites, apps, checkout flows. Agentic commerce is designed around intent specification. The customer no longer navigates; they delegate. The quality of the commercial relationship depends not on the usability of the interface but on the precision of the outcome specification and the fidelity of the agent's interpretation.
Shift 02
From Persuasion to Performance
Traditional commerce invests heavily in persuasion - branding, emotional design, conversion optimisation. Machine customers performing agentic shopping are immune to persuasion. They evaluate on performance: price, reliability, API response time, contractual terms, and outcome delivery. The entire apparatus of emotional commerce - from hero images to urgency timers - becomes irrelevant when the buyer is an algorithm.
Shift 03
From Loyalty to Liquidity
Human customers exhibit loyalty - habitual purchasing, brand affinity, switching inertia. Machine customers exhibit liquidity. They continuously evaluate alternatives, switch providers at machine speed, and maintain no emotional attachment to any brand. The competitive moat in agentic commerce is not brand loyalty but trust architecture - the structural reliability that makes an agent choose you over alternatives.
Shift 04
From Transaction to Relationship
Paradoxically, while machine customers are less loyal in the traditional sense, they create deeper structural relationships. An agent that manages your finances doesn't make one-off purchases - it maintains ongoing relationships with multiple providers, negotiates terms continuously, and optimises across a portfolio of services. The relational arc between agent and provider becomes the fundamental unit of commercial design.
Shift 05
From Visibility to Observability
Traditional commerce is visible - the customer sees the product, the price, the checkout. Agentic commerce is invisible by default. The human principal may never see the transaction. The design challenge shifts from creating compelling visible experiences to engineering observable systems - systems that can be audited, understood, and governed even when no human is watching.
05 - Trust as Commercial Infrastructure
The Structural Material of Autonomous Commerce
In traditional commerce, trust is a feeling. A customer trusts a brand because of experience, reputation, or emotional resonance. In agentic commerce, trust is structural material - the engineered foundation upon which every autonomous transaction depends. Without it, nothing moves.
The trust architecture of agentic commerce operates at three levels. First, principal-agent trust: the human must trust their agent to act faithfully within delegated authority. This is the delegation design problem - how scope is defined, how boundaries are enforced, how violations are detected and corrected. Second, agent-provider trust: the agent must evaluate the trustworthiness of commercial counterparties. This is computational trust - assessed through API reliability, contractual compliance history, and outcome delivery metrics. Third, system trust: the entire ecosystem must maintain integrity - the regulatory frameworks, the dispute resolution mechanisms, the audit trails that allow human oversight of autonomous commercial activity.
In agentic commerce, trust is not a brand attribute. It is infrastructure. It is the load-bearing structure upon which every autonomous transaction rests.
Trust debt - the accumulated cost of deferred trust investment - becomes a measurable commercial liability. Organisations that fail to build trust architecture into their agentic commerce systems will find that machine customers simply route around them. Unlike human customers, who may tolerate friction out of habit, machine customers will switch providers in milliseconds when trust signals degrade.
06 - The Invisible Transaction
Commerce Without Presence
The most radical implication of agentic commerce is the invisible transaction - a commercial exchange that occurs without any human being present at the moment of execution. Your financial agent switches your energy provider at 3am. Your procurement agent renegotiates a supplier contract while you sleep. Your insurance agent adjusts your coverage in response to a life event it detected from your calendar.
This is not automation in the traditional sense. Automation executes predefined rules. Agentic commerce involves judgement - the agent evaluates options, weighs trade-offs, and makes decisions that its human principal has not explicitly pre-approved. The design challenge is not how to make these transactions visible (that would defeat the purpose of delegation) but how to make them observable - auditable, explainable, and reversible when necessary.
The A2UI (Agent-to-User Interface) becomes the critical design surface - not the interface where transactions happen, but the interface where humans review, understand, and govern what their agents have done. The absent state - the state in which the agent operates without human attention - becomes the primary use state of agentic commerce. Designing for absence is the defining challenge of this new discipline.
07 - Designing for Absent Buyers
The Experience When No One Is Watching
Traditional UX design assumes a present user. Every pattern in the design toolkit - navigation, feedback, confirmation dialogs, progress indicators - assumes someone is watching. Agentic commerce inverts this assumption. The buyer is absent. The seller may also be absent. The transaction is a conversation between machines, governed by human intent but executed without human presence.
This demands a new design vocabulary. Absent-state design - the discipline of designing experiences for the moments when no human is engaged - becomes the foundation of agentic commerce UX. The consent architecture must be robust enough to govern transactions that the human has not individually approved. The failure architecture must handle errors gracefully when there is no human available to intervene in real time.
The design patterns for absent buyers include: pre-delegation contracts (clear specifications of what the agent is authorised to do before it acts), post-execution summaries (concise, auditable records of what happened while the human was away), exception escalation protocols (when and how the agent interrupts the human for decisions that exceed its authority), and trust calibration signals (continuous indicators that the agent is operating within expected parameters).
08 - The Banking Imperative
Why Financial Services Will Lead the Agentic Commerce Revolution
Banking is the industry where agentic commerce will have its most profound and immediate impact. Financial services already operate in a world of delegated authority (power of attorney, standing orders, direct debits), regulated trust (FCA Consumer Duty, PSD2), and invisible transactions (algorithmic trading, automated payments). The infrastructure of delegation already exists. Agentic AI simply extends it.
The banking imperative is threefold. First, banks must design for machine customers as account holders - agents that manage finances, optimise portfolios, and execute transactions on behalf of human principals. Second, banks must become agent-ready providers - their products and services must be discoverable, evaluable, and transactable by autonomous agents through APIs, not just human-facing interfaces. Third, banks must build the trust infrastructure that makes agentic commerce safe - the identity verification, the transaction monitoring, the dispute resolution, and the regulatory compliance that govern autonomous financial activity.
Central to this trust infrastructure is a new regulatory imperative: Know Your Agent (KYA). Just as Know Your Customer (KYC) verifies the identity and risk profile of a human customer, KYA establishes the identity, authority, behavioural boundaries, and principal lineage of an autonomous agent acting on that customer's behalf. KYA encompasses four pillars: agent authentication (proving the agent is what it claims to be), mandate verification (confirming the agent operates within its delegated authority), behavioural fingerprinting (detecting anomalous patterns that suggest compromise or drift), and principal traceability (maintaining an auditable chain from agent action back to human intent). For banks preparing for identic AI - consumer-deployed agents that evaluate, negotiate, and transact with zero brand loyalty - KYA is not optional. It is the regulatory and architectural prerequisite for admitting machine customers into the banking relationship.
The bank that masters agentic commerce will not be the one with the best app. It will be the one with the most trustworthy agent infrastructure.
09 - The Regulatory Landscape
Governance in the Age of Autonomous Transactions
The regulatory frameworks that govern commerce were designed for human actors. Consumer protection law assumes a human consumer. Financial regulation assumes a human decision-maker. Contract law assumes human parties with the capacity to understand and consent to terms. Agentic commerce challenges every one of these assumptions.
The FCA's Consumer Duty - which requires firms to deliver good outcomes for retail customers - does not yet contemplate a world in which the "customer" is an autonomous agent. GDPR's requirements for informed consent become architecturally complex when the data subject's agent is making decisions about data sharing on their behalf. The Consumer Rights Act's provisions for unfair contract terms must be reinterpreted when the entity agreeing to terms is an algorithm that has read and evaluated every clause in milliseconds.
This is not a problem for regulators alone. It is a design problem. The organisations that build agentic commerce systems must design regulatory compliance into the architecture - not as an afterthought but as a first-class design constraint. The consent horizon, the autonomous integrity framework, and the explainability standards that the AXD discipline defines are not optional enhancements. They are the minimum viable architecture for lawful agentic commerce.
10 - The AXD Commerce Framework
Twelve Frameworks for Designing Autonomous Commercial Systems
The AXD Institute's twelve practice frameworks provide the complete architectural toolkit for designing agentic commerce systems. Each framework addresses a specific phase of the agentic experience lifecycle - from the initial intent specification through active operation to post-execution audit. Together, they constitute the most comprehensive design methodology for autonomous commercial systems yet proposed.
| Phase | Framework |
|---|---|
| Pre-delegation | Intent Architecture |
| Delegation | Delegation Design |
| Active operation | Autonomy Gradient |
| Active operation | Trust Calibration |
| Active operation | Interrupt Pattern Library |
| Active operation | Multi-Agent Orchestration |
| Cross-session | Agent Memory |
| Post-execution | Absent-State Audit |
| Continuous | Explainability Standard |
| Recovery | Failure Architecture |
| Entry | Onboarding & Discovery |
| Constitutional | Ethical Constraint |
11 - The Commercial Horizon
The Future of Commerce Is Already Here
Agentic commerce is not a future state. It is a present reality in early deployment. Algorithmic trading already executes the majority of financial market transactions. Programmatic advertising already involves machine-to-machine negotiation for ad placement. Smart contracts already execute commercial agreements without human intervention. What is changing is the scope - from narrow, domain-specific automation to broad, general-purpose autonomous commercial agency.
The organisations that will thrive in this new commercial landscape are those that understand a fundamental truth: the design of agentic commerce is not a technology problem. It is a design discipline. It requires the same rigour, the same intentionality, and the same human-centred thinking that transformed industrial products into designed experiences in the twentieth century. The difference is that the "user" is now sometimes a machine - and the "experience" is often invisible.
The organisations that master agentic commerce will not be those with the most advanced AI. They will be those with the most intentional design.
This is why the AXD Institute exists. This is why Agentic Experience Design is not merely a specialisation within existing UX practice but a new discipline entirely. And this is why agentic commerce - the commercial expression of that discipline - will define the next era of economic design.
The trading floor is empty. The machines are transacting. The question is not whether this will happen. The question is whether we will design it well.
12 - AI Agents in Commerce
AI Agents in Commerce: The New Participants
AI agents in commerce are not simply tools that assist human buyers and sellers - they are emerging as autonomous participants in commercial ecosystems. These agents negotiate prices, evaluate product quality through structured data analysis, execute purchases within delegated authority boundaries, and manage ongoing supplier relationships without human intervention at the transaction level.
The introduction of AI agents in commerce creates a fundamental shift in market dynamics. When both buyer and seller may be represented by autonomous agents, the traditional assumptions of marketing, pricing, and customer relationship management no longer apply. Emotional branding gives way to verifiable performance data. Persuasive copy gives way to structured, machine-readable product specifications. The “customer journey” becomes an agent’s decision tree, optimised for utility rather than experience.
For organisations preparing for this shift, the AXD Institute’s frameworks provide the design foundations: trust architecture for establishing agent-to-agent confidence, delegation design for structuring human-to-agent authority, and the Four Pillars of Readiness for assessing organisational preparedness. The age of AI agents in commerce demands not just technological capability but intentional design.
13 - Agent-Mediated Commerce
Agent-Mediated Commerce: When Machines Transact for Humans
Agent-mediated commerce describes the commercial model in which autonomous AI agents serve as intermediaries between human principals and the marketplace. Unlike direct e-commerce (where humans browse and buy) or automated commerce (where rules-based systems execute predefined transactions), agent-mediated commerce involves intelligent agents that exercise judgement, negotiate terms, and make purchasing decisions within the boundaries of delegated authority.
The design challenges of agent-mediated commerce are distinct from both traditional e-commerce and B2B automation. In agent-mediated commerce, the human principal must be able to specify outcomes without specifying methods, trust the agent’s judgement in novel situations, and maintain meaningful oversight without micromanaging every transaction. This is the domain of delegation design - one of the core concerns of Agentic Experience Design.
Agent-mediated commerce is already emerging in sectors where transaction frequency is high and decision criteria are well-defined: grocery replenishment, subscription management, insurance comparison, and financial product selection. As agent capabilities expand, agent-mediated commerce will extend into higher-value, more complex transactions - requiring increasingly sophisticated trust architecture and autonomous shopping frameworks.