The Primary Material · AXD Institute
Trust in Agentic
Experience Design
In traditional UX, the designer works in attention. In AXD, the designer works in trust. Trust is not a feature of agentic systems - it is the material from which they are built.
"Trust is the only material that holds the relationship together when the human is absent and the agent acts alone. Every other design consideration - usability, efficiency, delight - is subordinate to this structural fact."
- AXD Founding Principle II: Trust is the Primary Material
Seven Dimensions
The Architecture of Trust
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.
From the Observatory
Trust in the Essays
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.
Trust Architecture
Designing the Structural Foundation of Human-Agent Relationships
Trust Recovery Protocol
Designing Systems That Heal
Trust Debt
The Hidden Liability in Every Agentic System
Temporal Trust
Designing Confidence Across Time Horizons
The Consent Horizon
Designing Permission for Systems That Never Stop Acting
Failure Architecture
Why Graceful Degradation is the Highest Form of Agentic Design
From the Practice
Trust Frameworks
Three of the twelve AXD Practice frameworks are directly concerned with trust design.
Extended Topics
Trust in AI Agents
Specialised explorations of trust in AI agents across specific domains and design challenges.
Agentic AI Trust
Trust calibration and architecture for autonomous AI systems
Agentic KYC
Trust verification and identity in agent-mediated finance
AI Agent Oversight
Monitoring, accountability, and human oversight of autonomous agents
AI Agent Payments Design
Trust architecture and delegation patterns for autonomous financial transactions
Frequently Asked
Trust in AXD
What is trust architecture in agentic AI?
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.
Why is trust more important than usability in agentic systems?
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.
How does trust differ from confidence 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.
What happens when trust fails in an agentic system?
Trust failure in agentic systems follows predictable erosion patterns: silent degradation (the agent underperforms without reporting it), expectation drift (the agent’s behaviour diverges from the user’s 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.
What is trust in AI agents and why does it matter?
Trust in AI agents is the structured confidence that a human principal places in an autonomous agent’s ability to act competently, consistently, and within delegated boundaries. Unlike trust in traditional software (which is binary — it works or it doesn’t), 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 — and without delegation, agentic commerce cannot function.