Let’s say we want some stepped gradients like the ones below, with a certain number of steps.

Chrome 139 is experimenting with Open UI’s proposed Interest Invoker API, which would be used to create tooltips, hover menus, hover cards, quick actions, and other types of UIs for showing more information with hover interactions. The Interest Invoker API makes these components declarative and hover-triggered, meaning that you create them with HTML, and then the web browser handles the mouseenter and mouseleave events for you, sans JavaScript.  

Before you dive deep into this week’s episode, here’s an important announcement on the future of Fundament we’d like you to read.

It’s time we stopped blaming the tools and started asking better questions about how we work, what we value, and how we make space for innovation again.

Summary:  UX in complex, specialized domains requires adapting familiar methods across the design lifecycle to address domain constraints and expert-user needs.

Artificial Intelligence (AI) has moved beyond a niche technology into a crucial business tool that can optimize operations, improve customer experience, and support data-driven decisions.

We’ve gone through glassmorphism, neumorphism, micro-interactions, and parallax scrolling. Some trends look amazing but add nothing. What’s a design trend you wish would just die already?

Despite promising results on synthetic benchmarks (e.g. Vending-Bench, SpreadsheetBench, DSBench), frontier models consistently underperform once they are deployed in complex, real-world situations.

Of all the current debates around AI, one critique has stayed with me: that it’s “flattening the bar.” Tools like ChatGPT, the argument goes, make everyone’s writing sound the same — generic, overly polished, stripped of nuance. The concern is real, and I share it.

Now we’re in the agentic era, and that billing octopus grew some new tentacles just for AI agent billing. Or is it a different octopus? I’m not sure.