Agentic Experience (AX) principles are not standardized yet like UX or HCI heuristics. They are emerging patterns drawn from AI-driven product design. Here’s a complete set that captures the current discourse:
Intent
Systems must understand and act on user goals, not just commands. The experience adapts to why the user is engaging, not just what they click.
Autonomy
Agents take initiative when appropriate. They perform tasks or suggest actions without explicit step-by-step instructions.
Proactivity
Anticipate needs and surface options before the user asks, based on context and patterns.
Collaboration
Treat the AI as a partner. Shared control and co-creation between human and agent.
Transparency
Show reasoning, data sources, and limitations clearly so trust can be built and maintained.
Adaptability
Learn from user behavior and adjust style, language, and workflow to the individual.
Context-awareness
Consider time, location, prior activity, and environment in shaping responses and actions.
Learning
Retain relevant history across interactions to provide continuity and personalization.
Multi-modality
Communicate across text, voice, visuals, and actions seamlessly.
Control and Oversight / Human In The Loop (HITL)
Users remain the final decision-makers. Provide override, undo, and feedback loops.
Safety and Alignment
Keep actions bounded by user consent, ethical limits, and risk controls.
Explainability
Agents should make their internal logic legible in plain terms to the end-user.
Discoverability
Users should easily see what the agent can do, with hints and affordances.
Seamless Handoff
When agents cannot complete a task, gracefully pass control back to the human or another system.
Scalability
Support growth from simple single-use agents to complex ecosystems without breaking consistency.
This list synthesizes academic research on HCI + AI, industry whitepapers, and ongoing design discussions. It is not fixed, new principles (like “delegation chains” or “collective agency”) are being debated.