Measuring Trust: Metrics for Trust Architecture in Agentic AI

What is Agentic Experience Design?

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.

How does AXD differ from traditional UX?

Why is trust architecture important for agentic AI?

Key concepts in Measuring Trust

How do measuring trust relate to agentic commerce?

  1. Agency requires intentional delegation — every agentic system begins with a designed act of delegation
  2. Trust is the primary material — AXD works in trust rather than attention
  3. Absence is the primary use state — the most consequential experiences happen when no one is watching
  4. Relationships have temporality — agentic experiences accumulate history over time
  5. Outcomes replace outputs — AXD designers specify results, not interfaces
DimensionTraditional UXAgentic Experience Design (AXD)
Primary materialAttention and affordanceTrust and delegation
User statePresent, navigatingAbsent, delegating
Design outputScreens and interfacesOutcomes and constraints
Temporal modelSession-basedRelationship-based
Success metricTask completionTrust calibration

Frequently Asked Questions

Can trust be reduced to a single number?

No. Trust is a multi-dimensional property that cannot be meaningfully captured by a single score. AXD measures trust across five behavioural indicators (delegation breadth, delegation depth, intervention frequency, recovery speed, and absence tolerance) and tracks these as trajectories over time. A single 'trust score' would obscure the structural information that designers need to improve the system.

What is the most reliable indicator of trust in an agentic system?

Recovery speed - how quickly the human re-delegates after a failure. This indicator captures the resilience of trust, which is more informative than the level of trust at any given moment. A system where users quickly re-delegate after failures has robust trust architecture. A system where users withdraw for extended periods has fragile trust architecture, regardless of how high trust was before the failure.

How does trust measurement differ from user satisfaction measurement?

User satisfaction is a declared attitude measured through surveys. Trust is a behavioural property measured through delegation patterns. A user can be satisfied with an agent (it does a good job) but not trust it (they still check every decision). Conversely, a user can trust an agent deeply (they delegate freely) while being mildly dissatisfied with specific outcomes. Trust and satisfaction are correlated but distinct - and for agentic systems, trust is the more consequential measure.

Key Takeaways

Trust is invisible. You cannot see it, weigh it, or count it. Yet it is the most consequential property of every human-agent relationship. The measurement problem in AXD is this: how do you quantify something that exists only in the relationship between a human and an agent, that changes moment to moment, and that the human themselves may not be able to articulate? Traditional approaches to trust measurement rely on surveys - asking humans to rate their trust on a scale. These approaches are inadequate for agentic systems for three reasons. First, they are AXD requires a different measurement paradigm - one built on The range of domains in which the human delegates to the agent. A human who delegates only grocery shopping trusts the agent less than a human who delegates grocery shopping, travel booking, and financial management. Increasing delegation breadth over time is a signal of growing trust. The consequence level at which the human delegates. A human who allows the agent to spend up to £20 autonomously trusts less than one who allows £500. Increasing delegation depth - higher spending limits, more complex decisions, higher-stakes negotiations - signals deepening trust. How often the human overrides, corrects, or checks the agent's decisions. High intervention frequency signals low trust - the human does not believe the agent can operate independently. Declining intervention frequency over time signals growing trust. A sudden spike in intervention frequency signals trust erosion. How quickly the human re-delegates after a failure. If the agent makes a mistake and the human immediately re-delegates (perhaps with tighter constraints), trust is resilient. If the human withdraws delegation for weeks or months, trust has been deeply damaged. Recovery speed is the most sensitive indicator of How long the human is comfortable leaving the agent to operate without checking in. A human who checks the agent's activity every hour trusts less than one who checks weekly. Inc

References and Citations

Gartner: Machine Customers Will Be a Multibillion-Dollar Opportunity Harvard Business Review: The Age of AI Agents McKinsey: The State of AI in 2024 About the AXD Institute Contact Us Email the AXD Institute Tony Wood on LinkedIn Tony Wood on X (Twitter)