AI UX DAILY
Sunday, July 19, 2026
3 stories · curated for designers
The stories
Today in AI Products
 Component Anatomy (Open Source Tool) |
Jul 17 |
New tool lets you annotate your live DOM instead of exporting screenshots from Figma
Component Anatomy is an open-source tool that lets you add data-part attributes directly to your rendered HTML components and get an interactive, synchronized anatomy panel alongside them. It works in plain HTML, Astro, or Storybook, meaning your component documentation stays in sync with the actual component rather than drifting from a stale screenshot. The tool is positioned as a direct replacement for the annotated screenshot workflow that most design teams still rely on.
| “ |
Swap your next round of annotated Figma exports for a Component Anatomy setup in Storybook, so your documentation reflects what's actually shipped rather than what was designed six months ago. — Designer's Takeaway |
 Google Vids |
Jul 17 |
Google Vids adds AI avatars so users can star in their own workplace videos
Google Vids now lets Workspace users create a digital avatar of themselves from a selfie and voice recording, then use it to generate videos from text prompts or reference images. The update also brings background swaps, lighting fixes, and step-by-step AI edits to the platform, moving it closer to tools like HeyGen and Synthesia. This is Google's clearest push yet to bring AI-generated presenter video into everyday business workflows, not just creative production.
| “ |
If your product includes any kind of video creation or async communication flow, audit whether your current UI anticipates AI-generated presenters as a first-class content type, since users will increasingly arrive with AI video as an output expectation. — Designer's Takeaway |
| Jakob Nielsen / UX Tigers |
Jul 17 |
Nielsen flags real usability problems in how AI 'skills' get discovered and reused
Jakob Nielsen's latest roundup includes a section on the usability of AI skills, the add-on capabilities that let AI assistants perform specific tasks like booking a flight or summarizing a document. Nielsen points out that users struggle to discover which skills exist, understand what they actually do, and reliably reuse ones that worked before. This is a concrete usability gap in the AI assistant category that designers building on top of these platforms are positioned to solve.
| “ |
Map the discovery and reuse flow for any AI skill or action your product exposes, and run a quick usability test asking users to find a skill they used before, since Nielsen's data suggests this is where the category is currently breaking down. — Designer's Takeaway |
Today's Idea
The gap between AI capability and AI usability is growing fast
Three separate stories this week point at the same problem from different angles: AI tools are adding capabilities faster than the interfaces around them can make those capabilities findable, trustworthy, or maintainable. Whether it's component documentation drifting from reality, AI avatars arriving without a clear UX model, or AI skills that users can't rediscover, the design work right now is less about shipping new AI features and more about making the ones that already exist actually work for people.
Stop shipping AI slop
Audit your AI design against 36 patterns
Drop a screenshot, get specific gaps and a Claude Code prompt to fix them. Free, no signup for the first audit.
Audit your design →