AI UX DAILY
Friday, September 25, 2026
4 stories · curated for designers
The stories
Today in AI Products
 Meta Muse |
Sep 24 |
Meta ships a Tamagotchi-style wearable to give Muse a physical home
At Connect 2026, Meta announced a small wearable device that runs Muse as a persistent, ambient AI agent. The device gives Muse a physical form factor separate from phones and glasses.
| “ |
Sketch what a persistent ambient agent looks like when it lives on the body: what it surfaces unprompted, when it stays silent, and how a user dismisses it. — Designer's Takeaway |
 Figma |
Sep 24 |
Figma publishes what it takes to ship quality products in the AI era
Figma's blog argues that reaching a prototype is now trivial, and that the real design challenge is sustaining velocity through to launch while making decisions that produce standout, differentiated products.
| “ |
Audit your current sprint: identify the one decision point where AI-assisted speed is causing your team to converge on the default option instead of the right one. — Designer's Takeaway |
| IxDF / Frank Spillers |
via TLDR Design |
IxDF publishes spatial UI guidelines with specific viewing angles and distances
UX consultant Frank Spillers outlines AR design rules: keep content within 30 degrees of central vision (hard limit at 50), place elements 1.25 to 5 meters away, favor direct hand gestures over menus, and design for short interruptible sessions.
| “ |
Apply the 30-degree and 1.25–5 meter rules as a checklist when reviewing any AR or spatial UI layout before it goes to prototype. — Designer's Takeaway |
| Wired / Instinct |
Sep 24 |
A real-money AI agent saved $550 and also wasted $64 in the same week
Wired tested Instinct, an AI agent platform fundraising at a $2.5B valuation, over a week of real tasks. It booked reservations, flagged phishing, and found savings, but also spent $64 on an erroneous action with no warning.
| “ |
Design a spend or action threshold that triggers a confirmation step before any agent action above a cost or consequence ceiling you define with your users. — Designer's Takeaway |
Today's Idea
AI agents need guardrails before they need features
This week's stories share a pattern: AI acting in the physical world and on real money exposes the gaps in how we communicate cost, consequence, and control to users. Design confirmation surfaces and action thresholds before your agent ships, not after the first $64 mistake.
Stop shipping AI slop
Turn your design into Claude skills
Drop a screenshot. See which of the 38 patterns you are missing and take them away as Claude Code skills. Free, no signup for the first audit.
Was this issue worth your time?