Agentic Experience Design (AXD) is the discipline for designing trust-governed relationships between humans and autonomous AI systems. Founded in September 2024 by Tony Wood in Manchester, United Kingdom, AXD addresses how humans delegate, calibrate, observe, interrupt, and recover trust in agentic AI.
| 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 |
Agentic AI refers to autonomous systems that pursue goals, make decisions, and take action on behalf of people - often without real-time human supervision. It is distinguished from generative AI by its capacity for goal-directed behaviour, environmental interaction, and autonomous operation.
Generative AI produces content on demand - text, images, code. Agentic AI goes further: it holds goals, makes intermediate decisions, and takes action in the world. The difference is between a system that answers and a system that acts. That shift creates entirely new design requirements around authority, trust, and recovery.
Examples include AI agents that book travel, manage investment portfolios, negotiate supplier contracts, triage medical referrals, and shop on behalf of consumers. In each case, the agent holds delegated authority and acts autonomously within defined constraints - the hallmark of agentic AI.
The primary risks are authority drift (agents exceeding their mandate), trust erosion (humans losing confidence in agent decisions), accountability gaps (unclear responsibility when agents fail), and compounding errors (autonomous actions that cascade before intervention). These risks are why Agentic Experience Design exists as a discipline.
The word 'agentic' describes a system that possesses Three properties distinguish agentic AI from other forms of artificial intelligence: An agentic system receives a goal (or infers one from context) and determines its own sequence of actions to achieve that goal. It is not executing a fixed script but reasoning about the best path forward. Agentic AI operates in the world - calling APIs, browsing the web, sending messages, making purchases, managing schedules. It does not merely produce text or images; it takes consequential action. The defining characteristic is the ability to operate without continuous human supervision. The human delegates, the agent acts, and the human may not re-engage until the task is complete or an exception arises. Generative AI and agentic AI are related but fundamentally different. This distinction matters because it changes the design challenge entirely. Generative AI requires prompt design and output evaluation. Agentic AI requires An agentic AI system typically consists of several interacting components: Agentic AI matters because it changes the fundamental relationship between businesses and their customers. When AI agents act as Machine customers do not browse. They query structured data, evaluate APIs, and compare machine-readable signals. Businesses that are not machine-readable will not be found. Machine customers evaluate trust through verifiable performance metrics, not brand sentiment or visual design. Reputation via Reliability replaces brand loyalty. Machine customers can negotiate, compare, and complete purchases in milliseconds. The entire purchase funnel collapses into a single autonomous decision. Every agentic AI system needs answers to questions that no existing discipline addresses: How does the human grant authority? How is trust calibrated over time? What happens when the agent fails while the human is absent? How does the relationship evolve? These are the questions AXD was built to answer.