On Friday, Anthropic Labs, the research branch behind Claude, introduced Claude Design, a new service that generates design systems, website prototypes, and slide decks.
It handles the early-stage visual work that typically requires a professional designer or tools like Canva and Figma, and is currently in a research preview.
Frederic Lardinois, Senior Editor for AI, writes in his story on the debut: How designers will use this remains to be seen, but Anthropic argues that it’ll be great for creating realistic prototypes, product mockups, pitch decks, and marketing collateral. It will also be for frontier design, where a designer may want to build code-powered prototypes and general design explorations that may otherwise be too time-consuming or cumbersome.
Chrome extension built on Holo3 model tests agents that can navigate websites and carry out tasks without integrations.
Most AI systems interact with software through structured pathways — APIs, code repositories, or tightly defined tool integrations. That setup works when systems are designed for automation, but much of the software people rely on every day isn’t. Internal dashboards, legacy tools, and web apps often require manual navigation, with no clean way for AI to plug in.
Live next week: The Next Step for CD — An API-Driven Platform
Continuous Delivery solved the deployment problem, but AI is outpacing the “tribal knowledge” needed to run today’s pipelines. Join us on April 23 to see how Internal Developer Portals transform complex delivery into seamless self-service. Learn to bridge the gap between infrastructure and developer ease by unifying catalogs, environments, and automated guardrails.
The script is well-known by this point. A major AI lab drops a new model. It makes fresh progress against well-known benchmarks. People start to use it. They say it’s amazing, but expensive. And then we repeat the argument when the nextAI lab drops their latest and greatest. But not this time.
New webinar: Your delivery pipeline’s impact on AI productivity
Organizations are rapidly pouring money into AI tools, expecting faster, more reliable software delivery—but gains aren’t showing up where it counts. In this session, we'll explore why AI gains stall, how pipelines absorb throughput, which Continuous Delivery capabilities matter most, and how to close the automation gap.
MCP servers turn Claude into a reasoning engine for your data
Learn how to build your first MCP server using TypeScript and connect it to Claude Desktop with this step-by-step calculator app tutorial for developers.
In this episode, we talk about why Microsoft built a fully managed service mesh that deliberately avoids the words “service mesh,” how Istio’s ambient mode makes that possible, and what AI workloads demand from the network layer.
