Deep dives: AI UX patterns, explained with demos.
Deep dives
Deep divesThe Eight States Every AI Feature Needs Before LaunchEmpty, thinking, streaming, partial, error, refusal, handoff, and limits: what each state must tell people, shipped examples, and how to test it before launch.Updated Oct 2, 2026 · 9 min
Deep divesHow to Design AI Citations & Sources UXA build playbook for grounded answers: Trust Scaffolding, citation pattern demos, and ChatGPT vs Perplexity teardown lessons.Updated Jul 14, 2026 · 12 min
Deep divesHow to Design AI Chat Composer UXA build playbook for AI intent capture: Chat UX patterns plus ChatGPT vs Claude vs Perplexity vs Gemini composer teardown lessons.Updated Jul 14, 2026 · 12 min
Deep divesHow to Design AI Output, Artifacts & Refinement UXA build playbook for keepable AI output: artifacts, refinement, and ChatGPT vs Claude vs Perplexity vs Gemini teardown lessons.Updated Jul 14, 2026 · 12 min
Deep divesHow to Design AI Memory & Personalization UXA build playbook for inspectable continuity: memory controls, persona settings, and ChatGPT vs Claude vs Perplexity teardown lessons.Updated Jul 14, 2026 · 12 min
Deep divesHow to Design AI Onboarding & First-Value UXA build playbook for time-to-first-success: example prompts, wizards, and Notion AI vs Lovable vs Manus teardown lessons.Updated Jul 14, 2026 · 12 min
Deep divesWhat is Human-in-the-Loop UX? A 2026 Guide with ChatGPT, Gemini & Perplexity ExamplesHuman-in-the-loop UX is the set of approval, edit, and handoff controls that keep a person in charge of consequential AI actions. The model can draft, plan, and prepare. A person owns send, spend, deploy, or delete.Updated Aug 18, 2026 · 12 min
Deep divesDesigning better AI chat (part 2 of 2)Citations, memory, thread navigation, export, and recovery patterns that make AI chat usable past the demo.Updated Apr 14, 2026 · 12 min
Deep divesDesigning better AI chat (part 1 of 2)Make AI chat clear and controllable: streaming, tool switching, drafts, and branching.Updated Apr 8, 2026 · 12 min- Deep divesClaude vs ChatGPT: composer design choicesHow ChatGPT and Claude's composers differ on tools, modes, and context, and what to reuse in your own product.Updated Dec 30, 2025 · 10 min
Deep divesThe trust stack: 7 patterns that make AI feel safe to useCapability is not adoption. Seven UX patterns that build trust in the interface: citations, process, uncertainty, and control.Updated Dec 15, 2025 · 8 min- Deep divesThe AI chat interface playbook: 5 patterns worth shippingSlash commands, @ mentions, follow-up chips, smart code blocks, and a regen carousel.Updated Dec 8, 2025 · 7 min
- Deep divesHow Cursor shaped AI coding UX: 5 patterns to stealArtifacts, Apply, branching, streaming with context, and tool use. Cursor wins on how AI sits in the edit loop.Updated Dec 1, 2025 · 7 min
- Deep divesDesigning for AI agents: 6 UX patternsWhen AI takes actions, show the plan, the tools, the queue, and the handoff.Updated Nov 24, 2025 · 8 min
- Deep divesVisual AI patterns from Midjourney, Figma, and RunwayVariation grids, inpainting, upscaling, style transfer, and controls beyond the prompt.Updated Nov 17, 2025 · 6 min
- Deep divesOnboarding users to AI: 4 patterns that shorten the blank promptUsers do not know what to ask. Example prompts, empty states, progressive unlock, and guided tasks.Updated Nov 10, 2025 · 6 min
Frequently asked questions
What is an AI UX deep dive?
Each deep dive explains a cluster of related interface patterns: how they work, when to use them, and what breaks in production. This hub includes 16 essays with embedded demos linked to the main pattern catalog.
How do deep dives relate to the pattern catalog?
Deep dives are narrative glue across multiple catalog patterns. A trust or chat essay connects five to seven patterns with product context; each demo links to a full pattern page with definition, anti-patterns, and examples.
Which deep dive should product teams read first?
Teams shipping chat should start with Designing better AI chat (parts 1 and 2). Teams adding autonomy or tools should read Designing for AI agents. Trust and safety work maps to The Trust Stack.