AI UX DAILY
Thursday, June 25, 2026
4 stories · curated for designers
The stories
Today in AI Products
| Jun 24 |
Figma's design agent now supports custom tools, real context, and skills
Figma updated its design agent to go beyond basic prompting. You can now give the agent custom tools, feed it real project context, and encode skills so that what lands on the canvas actually reflects your design system and style, not just a generic interpretation of a text prompt.
| “ |
Set up at least one custom tool or skill in Figma's agent this week and compare the output against a prompt-only result to see where context changes the quality of what gets generated. — Designer's Takeaway |
| Jun 24 |
Figma connects to Weave, bringing creative workflows alongside design frames
Figma announced a new integration with Weave, letting creative workflows live directly alongside Figma frames. The connection is designed to reduce the context switching between ideation and execution by keeping Weave assets and processes in the same workspace.
| “ |
Notice how co-locating creative workflows with design frames shortens the gap between inspiration and production, and consider auditing where your own team context-switches unnecessarily between tools today. — Designer's Takeaway |
| Jun 23 |
FIFA's custom scoreboard typeface fails at small sizes, exposing a real UI readability trap
FIFA's World Cup 2026 scoreboard uses a custom typeface called FWC26, designed by Alistair McCready. While it reads boldly at large display sizes, it breaks down at small sizes and low-resolution contexts, exactly where scoreboards appear in many broadcast and digital environments. The analysis on Pimp My Type breaks down why the letterforms cause legibility problems under real viewing conditions.
| “ |
Run your next custom or branded typeface through small-size and low-resolution stress tests before finalizing it, because a font that looks great in a presentation deck can fail badly in the contexts where users actually read it. — Designer's Takeaway |
| Jun 23 |
Priceline upgrades AI assistant Penny with Claude, sharpening its conversational travel planning
Priceline integrated Anthropic's Claude into Penny, its AI travel assistant. The upgrade is aimed at making Penny's responses more contextually aware and conversational when users are planning trips, comparing options, or asking follow-up questions across a multi-step booking flow.
| “ |
Study how Penny's multi-step booking conversation is structured now that it has a stronger model behind it, and consider whether your own conversational flows are designed to handle the follow-up questions users actually ask rather than just the first intent. — Designer's Takeaway |
Today's Idea
Context is the real design input for AI agents
Two separate Figma updates this week point at the same underlying shift: prompts alone are not enough. Custom tools, project context, and encoded skills are what turn a generic AI output into something that actually fits your product. That same principle applies to conversational AI in products like Penny, where the quality of the experience depends less on the model and more on how well the surrounding design accounts for what users need to say next.
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 →