Matt Shumer, an AI startup founder and investor, wrote what became the most-shared AI post of the past week. His argument: the gap between public perception of AI and what the latest models can actually do has become enormous. He describes his experience watching AI evolve from a "helpful coding assistant" to something that builds, tests, and ships complete applications with minimal human input. He then extends this beyond tech, arguing that law, finance, medicine, and most knowledge work are on a similar trajectory. It's a 19-minute read, and you may not agree with every prediction. However, I think it's worth reading in thinking about the pace of change with AI, however that ultimately manifests.

In this episode of Dive Club, Hannah Hearth (newly hired as Head of Product Design at Vercel) shares a candid look at how AI tools are reshaping design teams from the inside out. At Vercel, every designer uses Claude Code and Cursor in their daily work, not necessarily to ship every pixel, but to participate more directly in the shipping process, whether that's polishing the final 1%, building a prototype to give engineers a head start, or in Hannah's case, merging her first PR (pull request—a proposed change to the codebase) just weeks into the job.

The conversation goes well beyond tooling, though. Hannah describes a broader shift: design sprints shrinking from five days to one, leaders returning to individual contributor work to stay hands-on, and portfolios that now need to show experimentation and adaptability—not just polished case studies. Her advice for designers at any level: share your work early and often, invest in storytelling and buy-in skills, and treat this era of rapid change as an opportunity to learn, even when it's uncomfortable.

Designer Amelia Prasad shares her experience skipping Figma, Webflow, and AI design tool wrappers entirely, going straight from paper sketches to a live portfolio site using only Claude Code. She points out that most AI design tools don't actually eliminate the translation layer between design and code; they just move it. By describing her design intent directly to Claude Code and iterating in the browser, she moved from concept to live site faster than any traditional workflow. Responsive design, data visualization, and progressive disclosure were all built through conversation rather than tool-to-tool handoffs.

Her key takeaway: Claude Code is a multiplier for existing design skills, not a substitute for them. She's clear about when working in Claude Code currently works best, including personal projects and rapid prototypes.

My take is that now is the time for designers to get comfortable working in Claude Code. You'll be building a skill that matters more every day: articulating design intent in a way that AI can execute directly.

Over the past few weeks, stock prices for major SaaS companies — including Salesforce, Workday, Adobe, Shopify, and Monday.com — have dropped sharply. Monday.com fell more than 20% in a single day; Workday's CEO stepped down after a 17% decline in its stock price. The core fear driving the selloff: AI tools are becoming capable enough that companies may no longer need to buy seats for specialized software. When someone can throw their financial data into Claude and get useful analysis, or when a small team of developers can build and maintain internal tools that replace expensive per-seat subscriptions, the traditional SaaS business model starts to look vulnerable.

This past week, the Hard Fork podcast took a look at the so-called SaaSpocalypse—and reactions to it. Not everyone is panicking equally. Some argue that no one is going to "vibe code" their own payroll system, and that regulated industries like law and finance will still need robust, compliant software. And there is this argument: it's not that the technology will demolish everything overnight, but that it enables fundamentally different business models. Instead of buying 10,000 seats of HR software, SaaS companies may shift toward outcome-based pricing and leaner, AI-augmented workflows. The disruption is real, but it could be that it's more of a business model story than a pure technology story.

Thanks for reading, and see you next Wednesday with more curated AI/UX news and insights. 👋

All the best, Heidi