AI UX patterns
150+ curated AI UX patterns from shipped products, chat, agents, trust, onboarding, and more. Each entry includes when to use it, pitfalls, examples, and interactive demos where available.
Featured
Trust AI UX patterns
Prove sources, show reasoning, and make consent inspectable before users act.
Start here
Citations
Attach verifiable sources, quality signals, and claim previews
Chain of Thought
Reveal step-by-step reasoning behind an answer
Granular Consent
Per-capability permissions with scope, expiry, revoke, and drift
Failure Disclosure
Say clearly when the system cannot answer or a tool failed
Chatbot AI UX patterns
Ship turn-taking, streaming, repair, and memory before chat polish.
Start here
Agents AI UX patterns
Make tools, approval, budgets, and rollback visible before agents act.
Start here
Onboarding AI UX patterns
Guide a first success with starters, wizards, and progressive unlock.
Start here
Inputs AI UX patterns
Clarify tools, attachments, modes, and templates before the user sends.
Start here
Tool Switching in Composer
Switch between AI capabilities within composer
Multimodal Input
Combine images, files, and text in one turn
Context Chip Management
Adding context sources via menu with removable chips
Prompt Starters
Example prompts for empty states, templates, and libraries
Outputs AI UX patterns
Stream answers, refine drafts, compare options, and keep feedback tight.
Start here
Design Tools AI UX patterns
Generate layouts, variants, and canvas edits users can compare and refine.
Start here
Commerce AI UX patterns
Find by meaning or image, cart across merchants, compare, and complete the buy.
Start here
Collab AI UX patterns
Share sessions, show presence, and review AI changes before they stick.
Start here
Audio AI UX patterns
Show listening state, live transcript, and safe interrupt for voice turns.
Start here
Performance AI UX patterns
Show model choice, cost, wait time, and limits before users hit a wall.
Start here
Frequently asked questions
What counts as an AI UX pattern here?
Each pattern is a named interaction or UI convention for AI products, streaming replies, tool pickers, citation chips, autonomy budgets, and similar, with a short definition, when to use, when not to use, anti-patterns, and product examples. High-intent pattern pages also include a product implementation table and a per-page FAQ.
How many patterns are in the catalog?
The browse catalog lists 150 patterns organized across 11 categories (chatbot, agents, trust, onboarding, commerce, inputs, outputs, and more). New patterns are added as products ship new conventions; category filters and search help you narrow the grid.
How do I use these patterns in product work?
Start from the user risk: trust, cost, reversibility, or latency. Pick a category filter or search by product name, open 2-3 similar patterns, compare anti-patterns, then link the pattern URL in your spec or Figma brief. Pair with Frameworks when you need vocabulary for agentic or chat territories.
Can I filter patterns by AI product or company?
Yes. Use search on this page or scan the examples listed on each card. Patterns reference products like ChatGPT, Claude, Perplexity, Cursor, and others where that convention appears in production.
How do patterns relate to Frameworks?
Frameworks define territories and guiding questions for a whole problem space (for example permission, rollback, and autonomy for agents). Patterns are the catalog entries you implement; framework pages map which patterns belong to each territory and which anti-patterns to avoid.
Are interactive demos available for every pattern?
Most pattern detail pages include a live demo illustrating states and flows. If a pattern is conceptual or documentation-only, the page still lists real-world examples from shipped products so you can research behavior in context.