Agentic UX (AX) Research Summary

This document captures emerging UX patterns for agent-based platforms, with a focus on enterprise use cases in AI-driven marketing and business intelligence.

Core Shift: From UI to Agentic Experience

Agent-based systems are driving a shift from command/response UX to delegation UX:

• Users express intent and goals.
• Agents decompose and execute multi-step workflows.
• Interfaces provide transparency, supervision, and governance controls.

Emerging UX Patterns

1) Conversational and Voice-Driven Interaction

Natural language is increasingly the primary interaction model for AI products.

Patterns - Intent-driven task specification (goal > screen-by-screen navigation) - Multimodal interaction (text + voice + visuals) - Adaptive prompts and follow-ups based on context

Design implications - Conversation becomes a primary navigation layer. - Visual UI supports reasoning and state awareness. - Voice enables fast, hands-free task exploration.

2) Real-Time Agent Transparency

As autonomous behavior increases, visibility becomes mandatory for trust.

Patterns - Live status indicators (what the agent is doing now) - Intent summaries (why an action was selected) - Action histories and audit logs (what changed, and when) - Explainability on demand (detail when needed, not always) - Human-in-the-loop interruption/override controls

Enterprise impact - Teams can verify optimization changes before accepting them. - Decision rationale can be inspected for compliance and accountability.

3) Task-Resolution Visualization

Agent systems run multi-step plans, so linear chat alone is often insufficient.

Patterns - Workflow graphs of actions and dependencies - Planning previews before execution - Timelines and checkpoints for long-running tasks - Role-aware visualization for multi-agent coordination - Node-tree exploration of context and reasoning paths

Benefit - Improves comprehension and user trust. - Makes it easier to diagnose and correct workflow failures.

4) Trust-Oriented Design Systems

Agentic products require design system primitives for autonomy.

Components - Explainability modules - Confidence and uncertainty indicators - Delegation controls (levels of autonomy) - Pause/resume/interrupt controls - Governance dashboards and logs

Architecture trend - Agent-compatible semantic UI components can improve automation reliability and observability versus brittle interaction with unstructured interfaces.

Convergence Themes

Across current literature and practice, four themes are converging:

1. Intent replaces navigation
2. Transparency becomes part of the interface
3. Workflows replace static dashboards
4. Human oversight remains central

Strategic Value

AI-Driven Marketing

• Conversational campaign planning
• Autonomous optimization with visible reasoning
• Workflow maps for campaign operations
• Lower operational burden with maintained strategic oversight

Business Intelligence

• Conversational data exploration
• Automated analysis workflows
• Explainable insights with reasoning paths
• Transition from static dashboards to collaborative intelligence

Recommended UX Layering for Agent-Based Products

A practical AX architecture commonly includes:

1. Intent interface (conversation/voice)
2. Transparency layer (status, rationale, logs)
3. Workflow visualization layer (plans, progress, dependencies)
4. Governance layer (approval, policy, audit)

Sources

1. https://medium.com/%40birdzhanhasan_26235/ai-ux-design-patterns-research-ff7b8056d07d
2. https://uxmag.com/articles/designing-for-autonomy-ux-principles-for-agentic-ai-systems
3. https://medium.com/%40pro.namratapanchal/what-are-the-must-know-agentive-design-patterns-for-2026-21cf34839a01
4. https://arxiv.org/abs/2505.19101
5. https://arxiv.org/abs/2510.06457
6. https://www.uxmatters.com/mt/archives/2025/12/designing-for-autonomy-ux-principles-for-agentic-ai.php
7. https://arxiv.org/abs/2601.14790