AI UX DAILY
Saturday, April 25, 2026
4 stories · curated for designers
The stories
Today in AI Products
| Apr 24 |
10 Guidelines for Designing Your Site's AI Chatbots
Nielsen Norman published research-backed guidelines for site-specific AI chatbots that actually help users. Key findings: effective chatbots clearly state their capabilities upfront, offer relevant prompt suggestions based on page context, and quickly signal what they know about the current screen. This moves past generic chatbot patterns to context-aware design.
| “ |
Audit your chatbot's first message: does it tell users what it can and cannot do? Add capability statements and contextual prompt suggestions to your chat UI within the next sprint. — Designer's Takeaway |
| Apr 24 |
ChatGPT Image Generation Now Integrated Directly in Figma
Figma added native ChatGPT image generation to its canvas, letting designers generate and iterate on visual assets without leaving the design tool. The integration includes one-click prompt workflows and image-to-canvas placement.
| “ |
Notice how Figma embedded AI generation as a canvas tool rather than a separate panel—this reduces context switching. Consider where your design workflow has forced tool-switching and whether embedding AI capabilities inline would reduce friction. — Designer's Takeaway |
| Apr 24 |
Nothing Launches On-Device AI Dictation Tool Supporting 100+ Languages
Nothing released an on-device dictation tool powered by AI that processes speech recognition locally, supporting over 100 languages without requiring cloud transmission. The tool integrates into the OS input layer and handles accents and context better than traditional speech-to-text.
| “ |
On-device AI changes input design assumptions: you can now build richer, faster voice input experiences without network latency concerns. Explore voice as a first-class input method in your next project, not just accessibility scaffolding. — Designer's Takeaway |
| Apr 24 |
Instagram Instants: AI-Powered Quick Photo Sharing
Instagram launched Instants, a feature that uses AI to suggest auto-enhance and auto-caption options for photos before sharing. The feature uses computer vision to detect context (food, people, places) and auto-generate contextual captions users can edit or accept.
| “ |
See how Instagram turned AI suggestions into optional defaults rather than required workflows. Apply this pattern: offer AI-generated content as a starting point users can refine, not replace, to preserve agency while accelerating the creation process. — Designer's Takeaway |
Today's Idea
Context and Capability Transparency Are Non-Negotiable
This week's releases show a consistent pattern: AI features that work best for users are ones that explain what they can do, show their work in real-time, and preserve user control over the final output. Whether it's Nielsen Norman's chatbot guidelines, Figma's integrated image generation, or Instagram's optional auto-captions, the common thread is that AI should propose, not dictate. Designers should prioritize capability statements and default-to-edit workflows over black-box automation.