Anthropic shipped a new product today called Claude Design, and it is a direct shot at the way visual work gets made. You describe what you want, Claude builds a first version, and you refine it through conversation, inline comments, or custom sliders it generates on the fly. When it looks right, you export to Canva, PDF, or PowerPoint, or hand it straight to Claude Code to turn into a real product.

The pitch is simple. Designers never have enough time to explore more than a handful of directions. Founders and product managers have ideas they cannot execute because they do not have a design background. Claude Design tries to solve both problems at once. Early users at Datadog say what used to take a week of back-and-forth now happens in a single conversation. Brilliant says their most complex pages, which previously took 20 prompts in other tools, needed two in Claude Design.

Figma, the industry standard design tool most professional teams rely on for everything from mockups to developer handoff, has been coming up a lot in discussion today. Claude Design is not a replacement for that, at least not yet. What it does is make the earliest, messiest part of the design process faster and more accessible. Whether that eventually bleeds into Figma's territory or simply expands the market is genuinely unclear.

The more interesting question is what happens if Claude Design gets really good at the parts Figma owns. That is not today's problem, but it is worth watching.

Kevin Weil, Bill Peebles, and Srinivas Narayanan all announced they were leaving OpenAI today. Weil ran the science division. Peebles built Sora from the ground up. Narayanan led B2B applications. Three departures, one day, no warning.

Narayanan's exit is reportedly personal. The other two are harder to explain away. OpenAI has spent the past month making it very clear that it is narrowing its focus, cutting products that do not feed directly into its core bets on coding, enterprise, and frontier research. Sora was the first product to get shelved. Now the science team is being dissolved into other groups, and Prism, a standalone AI workspace built specifically for researchers, is being absorbed into Codex. The pattern is obvious: if it does not serve the main mission, it is gone.

Peebles built one of the most impressive AI products of the last two years and leaves with nothing but kind words. Weil spent two years going from Chief Product Officer to leading a research team and ends up watching that team get reorganized out of existence. Both exits look clean on the surface. The products underneath tell a different story.

The deeper question is whether folding science AI into a developer tool is actually the right call. OpenAI is betting that researchers will meet them where Codex lives. That is a big assumption about who their users are and what they need. With Spud reportedly close, they are betting on a very tight org to land it. Today did not make that look easy.

Google Flow Music (formerly called ProducerAI) launched today as a standalone site where you can create, share, and remix original music using plain text prompts. Describe what you want and it produces a fully arranged track. It sits alongside Google Flow, which handles images and videos, giving Google a creative suite that now covers music, video, and visuals all under one roof.

It is a quieter launch compared to everything else today, but the direction is clear. Every major creative medium is getting an AI native tool, and Google is making sure it has a seat at that table.