AI & Agentic UX
What Is Agentic UX? A Practical Guide for Product Designers
Agentic UX is the design of experiences in which a system carries out work toward a user's goal. It covers delegation, scope, visibility, intervention, and…
OSAMA ALI / FIELD NOTES
AI & Agentic UX
Explore practical AI & Agentic UX guides, examples, and implementation advice.
20 articles to explore
Search could not load. You can still browse the articles below.
AI & Agentic UX
20 articles
AI & Agentic UX
Undo, Approve or Confirm? Designing Controls for AI Agents
Use undo for reversible work, approval before consequential delegated actions, and confirmation when the user needs to verify a specific commitment. Repeated…
AI & Agentic UX
The 8 UI States Every AI Feature Needs
Plan for idle, input collection, processing, partial output, completed output, uncertainty, failure, and recovery. These are a design checklist, not a universal…
AI & Agentic UX
Prompt UX: How to Design Better Inputs for AI Products
Help users express the task, relevant context, constraints, and desired output without requiring them to master prompt jargon. Examples, editable starting…
AI & Agentic UX
Human-in-the-Loop UX: When Should AI Ask for Approval?
Ask for approval when an action is consequential, difficult to reverse, or outside the user's established permission. Put the actual proposed change and its…
AI & Agentic UX
How to Measure the UX of an AI Agent
Measure whether delegated tasks finish correctly, how much supervision they require, and how costly failures are to recover from. Pair completion and correction…
AI & Agentic UX
How to Design Loading and Waiting States for AI Products
Show which stage of work is happening and whether the user can continue, cancel, or return later. Use a time estimate only when the system can support it;…
AI & Agentic UX
Designing UX for Non-Deterministic AI Systems
Plan for a range of possible outputs rather than one ideal response. Keep interaction rules predictable even when the generated content varies: users still need…
AI & Agentic UX
Designing AI Onboarding: Teaching Users What AI Can and Cannot Do
Teach the first useful task with a realistic example, then explain the limits at the moment they matter. Users should understand required inputs, expected…
AI & Agentic UX
Designing AI Citations and Sources Users Can Trust
Connect each important claim to the specific source that supports it. Make the source easy to open, distinguish retrieved evidence from generated interpretation,…
AI & Agentic UX
Designing AI Agents: UX Patterns for Products That Act, Not Just Chat
Start with the agent's authority, not its chat box. Define allowed actions, permission boundaries, approval points, and recovery paths before designing the…
AI & Agentic UX
Conversational UX vs Agentic UX: What's the Difference?
Conversational UX focuses on an exchange of messages. Agentic UX also handles delegated work: what the system may access, which actions it may take, how progress…
AI & Agentic UX
AI UX Design: The Complete Guide to Designing AI Products
Design AI around a useful task, a clear boundary of responsibility, and recoverable outcomes. The interface must explain what the system can do, show when it is…
AI & Agentic UX
AI Trust UX: How to Design Products Users Can Actually Trust
Help people judge when to rely on an AI feature. Show its scope and limitations, connect important claims to evidence, and make correction and recovery available…
AI & Agentic UX
AI Permissions UX: Designing Safe Agent Access and Actions
Request access when its purpose becomes clear, limit it to the task, and show how to revoke it. Separate permission to read information from permission to change…
AI & Agentic UX
AI Memory UX: How Should Products Remember Users?
Make remembered information visible, editable, and removable. Distinguish information used in this conversation from information retained across sessions, and…
AI & Agentic UX
AI Explainability UX: How Much Should the Product Explain?
Explain enough for the user to judge the next decision: what information was used, what limits apply, and how to challenge the result. Offer deeper detail on…
AI & Agentic UX
AI Error Recovery: Designing When the Model Gets It Wrong
An AI error needs a recovery path, not just an apology. Preserve the user's work, identify which outputs or actions are affected, and offer a specific…
AI & Agentic UX
AI Confidence and Uncertainty: How Should Interfaces Communicate It?
Explain what is known, what remains uncertain, and what the user can do next. A confidence percentage is useful only when its meaning is understood and its…
AI & Agentic UX
Agentic UX Patterns: Delegation, Approval, Intervention and Recovery
Give users a clear handoff: define the goal, preview consequential actions, show progress, and provide a way to interrupt or recover. The right control depends…
No matching articles
Try a broader phrase or clear your filters.
READ WITH A PURPOSE
A useful place to start.
Follow a short sequence around the problem you are working on.