How trust is built, maintained, and eroded over time through consistent agent behaviour. Designing for the long arc of human-agent relationships..
| 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 |
Temporal trust recognises that trust in agentic systems is not static but changes over time. It accumulates through positive interactions, decays during periods of inactivity, and can be destroyed by failures. In AXD, temporal trust is a design material - designers must model how trust evolves and design agent behaviour that responds to the current trust state.
Trust decay occurs when an agent has not recently demonstrated competence or reliability. Even without failures, prolonged absence of positive evidence causes trust to erode. This means agents must periodically demonstrate their value and reliability to maintain trust levels. The rate of decay depends on the relationship maturity and the stakes involved.
Temporal trust and the consent horizon are deeply connected. As trust decays over time, the validity of the original consent also diminishes. The consent horizon marks the point where trust has decayed sufficiently that the agent should seek re-authorisation. Designing these two concepts together ensures that agent authority remains aligned with current trust levels.
Temporal trust recognises that trust in agentic systems is not static but changes over time. It accumulates through positive interactions, decays during periods of inactivity, and can be destroyed by failures. In AXD, temporal trust is a design material - designers must model how trust evolves and design agent behaviour that responds to the current trust state.
Trust decay occurs when an agent has not recently demonstrated competence or reliability. Even without failures, prolonged absence of positive evidence causes trust to erode. This means agents must periodically demonstrate their value and reliability to maintain trust levels. The rate of decay depends on the relationship maturity and the stakes involved.
Trust Built Over the Long Arc of a Relationship Trust built, maintained, or eroded over time through consistent agent behaviour. Temporal Trust posits that every interaction with an agent is a single data point on a continuous timeline. A solitary successful transaction might instill a fleeting sense of reliability, but it is the consistent, predictable, and coherent behavior of an agent over weeks, months, and even years that forges a deep and abiding sense of partnership. It is the agent that remembers your preferences from a conversation six months ago, the one that anticipates your needs based on a pattern of behavior it has observed over time, the one whose core personality remains stable even as its capabilities expand. This long-term perspective forces us to design not for single moments of delight, but for a sustained and meaningful connection. It is a commitment to building a legacy of trust, one interaction at a time. The architecture of temporal trust - a core concern of Beyond performance, enduring trust requires Of all the pillars supporting temporal trust, consistency is arguably the most important and the most challenging to achieve. From a psychological perspective, consistency breeds familiarity, and familiarity breeds trust. Humans are pattern-matching creatures; we find comfort and safety in predictable systems. When an agent behaves in a consistent manner, we can build an accurate mental model of its "personality" and its likely responses. This mental model reduces cognitive load and allows for a more fluid and intuitive interaction. We no longer have to second-guess the agent’s intentions or waste mental energy trying to understand its erratic behavior. The agent becomes a known quantity, a reliable partner in our digital lives. The corrosive effect of inconsistency can be subtle but profound. Imagine an agent that is cheerful and helpful one day, then curt and dismissive the next. Or an agent that suddenly changes its core interface or decision