Trust is the primary material of Agentic Experience Design (AXD). Explore trust architecture, trust erosion patterns, trust signals, and trust measurement in agentic AI and agentic commerce..
| 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 |
Trust architecture is the structural design of confidence in autonomous AI systems. It encompasses the four layers of trust - predictability, agency, communication, and evolution - that together form the load-bearing structure of every human-agent relationship. Trust architecture is the primary design discipline within AXD (Agentic Experience Design), replacing attention as the core material that designers work with.
In traditional software, the user is present and navigating an interface - usability determines whether they can complete a task. In agentic systems, the user is absent and the agent acts autonomously. The question is no longer 'can the user complete the task?' but 'does the user trust the agent to complete the task on their behalf?' Trust governs delegation, and delegation governs everything in agentic commerce.
Confidence is a momentary state - a snapshot of how the user feels about the agent right now. Trust is a structural property - the accumulated history of competence, consistency, and recovery that determines whether the user will delegate again tomorrow. AXD designs for trust, not confidence, because agentic relationships are temporal: they accumulate history, evolve through failure, and deepen through demonstrated reliability over time.
Trust failure in agentic systems follows predictable erosion patterns: silent degradation (the agent underperforms without reporting it), expectation drift (the agent\u2019s behaviour diverges from the user\u2019s mental model), catastrophic breach (a single high-consequence failure that collapses accumulated trust), and recovery stall (the system lacks mechanisms to rebuild trust after failure). AXD provides design frameworks for detecting, preventing, and recovering from each pattern.
Trust in AI agents is the structured confidence that a human principal places in an autonomous agent\u2019s ability to act competently, consistently, and within delegated boundaries. Unlike trust in traditional software (which is binary \u2014 it works or it doesn\u2019t), trust in AI agents is graduated, contextual, and temporal. It must be designed, calibrated, and maintained through intentional trust architecture. Without trust in AI agents, delegation cannot occur \u2014 and without delegati
The Primary Material · AXD Institute Trust in agentic systems is not a single phenomenon. It is a composite architecture with distinct dimensions - each requiring its own design language, its own failure modes, and its own measurement framework. These seven pages constitute the AXD Institute's canonical treatment of trust. Trust architecture is a recurring theme across the Observatory's forty-seven long-form essays. These six are the most directly concerned with trust as a design material. Three of the twelve AXD Practice frameworks are directly concerned with trust design.