Framework

10 Heuristics for
Agentic Experience

Inspired by Nielsen Norman Group usability heuristics — reimagined for agentic systems, AI agents, and multi-agent environments.

What makes these different

Traditional UX heuristics optimize for clarity of interface.

AX/AUX heuristics optimize for quality of relationship.

The shift: from interaction design to relationship design.

Heuristic 01

Visibility of Agent Intent & Action

The agent should always make its intent legible before or during action.

Users shouldn't guess why the agent is doing something. Use patterns like:

  • Intent playback
  • Plan preview
  • "Here's what I'm about to do…"
Maps to: Intent Handshake (U2)
Heuristic 02

Progressive Transparency (Not Full Transparency)

Agents should explain themselves proportionally to trust and context.
  • Early: full reasoning, sources, confidence
  • Later: concise or silent execution

This is core to AX: Transparency → Summary → Confident → Silent

Core to: AX transparency model
Heuristic 03

User Control Through Steering, Not Micromanagement

Users guide agents via intent and correction — not step-by-step control.

Replace “Click here → do this → confirm that” with:

  • Approve / modify / redirect

Key shift: UX control → AUX steering

Key shift: UX control → AUX steering
Heuristic 04

Trust is Dynamic, Not Static

Agent behavior must evolve as trust develops over time.
  • New user: cautious, gated
  • Experienced user: autonomous, fast

Modeled via trust timeline and progressive stages.

Modeled via: Trust Timeline / stages
Heuristic 05

Clear Boundaries of Autonomy

Users must always understand what the agent can and cannot do.

At any moment, users should know what is:

  • Autonomous — agent acts freely
  • Confirmed — agent asks first
  • Blocked — agent cannot act
From: Trust topology + autonomy levels
Heuristic 06

Graceful Handling of Uncertainty & Failure

Agents must surface uncertainty and degrade gracefully.

Instead of hallucinating:

  • Ask
  • Escalate
  • Provide partial output

Includes confidence thresholds, escalation paths, and failure design as first-class.

Includes: Confidence thresholds & escalation paths
Heuristic 07

Appropriate Agent Assertiveness

Agents should challenge users when necessary — not blindly comply.

Assertiveness levels:

  • Compliant → Advisory → Assertive → Protective

“Pushback is professional.”

Principle: Pushback is professional
Heuristic 08

Context Efficiency & Awareness

Agents must use context intelligently, not exhaustively.

Avoid full scans and context overloading. Prefer:

  • Routing
  • Indexing
  • Summaries

Core AUX problem: context = cost + performance constraint.

Core problem: Context = cost + performance constraint
Heuristic 09

Multi-Actor & Multi-Agent Clarity

Systems must clearly show who (or what) is acting at any moment.

Human vs agent vs agent. In multi-agent systems:

  • Who delegated?
  • Who owns the outcome?
From: Multi-agent choreography & Org Map topology
Heuristic 10

Consistency of Behavior, Not Interface

Agents must behave consistently — even if the interface changes.

Unlike classic UX, the UI can adapt (generative UI). But these must remain stable:

  • Decision logic
  • Trust boundaries
  • Tone
Critical for: Adaptive interfaces (U7) & generative UI systems

Apply these heuristics
to your product

We help teams design agentic experiences that people actually trust. Let's talk about your next build.

Start a Project →