AI for Designers
How designers ship with AI. Tools, workflows, prompts, and the practical craft of building AI-powered products.
All AI for Designers articles
Claude Effort Levels: From Low to Ultracode
A plain-English guide to Claude's /effort levels for designers and creators, from low and medium to xhigh, max, ultracode, and auto, with a simple rule for which to use when.
What Claude Opus 4.8 Means for Designers
A designer's teardown of Claude Opus 4.8. The 1M context window, fast mode in Claude Code, the long-horizon reliability gains, and the four design workflows it actually changes. Plus an honest list of what Opus 4.8 still cannot do for you.
Google Stitch for Designers: How Prompt-to-UI Actually Works in 2026
What Google Stitch does for designers in 2026, the workflows that pay off, and where it loses to v0, Lovable, and Figma Make. Real outputs, honest verdict.
Cursor for Designers: The 2026 Guide to Actually Useful AI in Your Design Workflow
What Cursor actually does for designers in 2026, the real workflows that pay off (design system maintenance, prototype rewrites, MCP-driven Figma to code), and where it still falls short.
MCP Servers for Designers Worth Installing in 2026
The MCP servers worth installing if you work in Figma, Cursor, or Claude. What MCP is, which five servers ship real daily value, and how to wire them up.
GuideDesigning Agent Memory: The 2026 Designer's Handbook
Agent memory is the new AI design surface no one teaches. Build memory features users actually trust with 4 types, 5 trust principles, plus a workshop.
GuideGenerative UI Design: The Practical Handbook for 2026
Generative UI design explained: the four architectures, the pattern language, the failure modes, and the practical handbook for designers shipping in 2026.
Designing Streaming UIs
The streaming output region is the new canvas. A working playbook for designing AI streaming UIs as a real interaction model, with five layers, real product teardowns, anti-patterns, and a pre-ship audit.
Designing for Agents Watching
AI agents are now first-class users of your product. The 2026 design constraint: every surface needs a human plane and a machine plane, or you lose the next agent integration.
Onboarding Without Onboarding
The 5-step welcome modal is dead. The five patterns AI-native products use to put new users inside real work in under thirty seconds, and the cases where onboarding still earns its keep.
Prompt Surfaces: The New UX Primitive
The prompt input is the new button. A working playbook for designing prompt surfaces as a first-class UX primitive, with anatomy, patterns, failure modes, and a pre-ship audit.
Designing Trust: How AI Products Win or Lose User Confidence in the First 5 Minutes
A working playbook for designing trust into AI products. Real teardowns of Claude.ai, Cursor, Granola, Perplexity, Linear AI, ChatGPT, and Notion AI. Six trust patterns that earn confidence in the first five minutes, four anti-patterns that destroy it, and a five-bullet checklist any designer can run on an AI surface tomorrow.
The MCP Era: How Model Context Protocol Quietly Reshaped AI Apps in 2026
A working primer and state-of-the-protocol on Model Context Protocol heading into mid-2026. What MCP actually is at the wire, why it won where prior agent-tool standards failed, the canonical servers shipping in production, what designers and builders should do with it, and where MCP still loses.
The Computer Use Era: When AI Agents Can Actually Run Your Software
A working playbook on AI computer use heading into mid-2026. What Anthropic Computer Use, OpenAI Operator, and browser-native agents actually do, where they ship, where they still break, and the design and dev decisions every team needs to make before the agents start using their product.
The 2026 Frontier Model Map: GPT-5.5, Claude 4.7, Gemini 3, and What Each Does Best
A working map of the 2026 frontier model landscape. GPT-5.5, Claude 4.7 Opus and Sonnet, Gemini 3 Pro, Llama 5, Grok 4, DeepSeek V4, and Qwen 3 graded on what they actually win at, where they leave money, ballpark pricing per million tokens, and a decision matrix for designers and builders picking models for real product stacks.
What Claude 4.7 Actually Changed for AI Builders
A working teardown of Claude 4.7 for AI builders. Agent reliability past two hours, 1M context standard across the family, computer use generally available, prompt caching tier improvements, and the Sonnet and Haiku speed jumps that opened high-throughput workloads.
Designing for Latency: When AI Speed Becomes UX
A working playbook for designing around AI latency. Streaming text, optimistic UI, progressive disclosure, reasoning surfaces, and background agents, with real teardowns of Claude.ai, Cursor, Linear AI, Granola, and Perplexity. Plus the math of perceived speed.
Prompts as Components: How Designers Build Reusable Prompt Libraries in 2026
Components made design scalable in the 2010s. In 2026, prompts are the new components. A working playbook for designers building reusable prompt libraries: anatomy, variants, versioning, distribution, and the new prompt librarian role.