Unbiased AI-powered news
The Instagram co-founder, who joined Anthropic in May 2024 when its product and engineering organization numbered about 30 people, moved into the company’s internal moonshot group four months ago. Krieger told Alex Heath at the firm’s developer conference last week that the current separation between Claude AI, Claude Code and Cowork is a broken abstraction.
pymnts.comMike Krieger moved into Anthropic’s Labs team about four months before May 15, 2026, after telling the company’s co-founder and president Danielle Amodei last fall that he wanted to leave to start a new company. By then the product organization had grown from about 30 people when Krieger joined in May 2024 as chief product officer to several hundred.
He ended up staying to run a new incarnation of the Labs team.
Labs is the internal moonshot group at Anthropic that produced Claude Code, MCP, Agent Skills and Computer Use. The team now runs on two-week sprints with mostly former founders. ” Krieger told Alex Heath at Anthropic’s developer conference last week that Anthropic is “shipping our harness strategy” rather than a product.
Krieger does not think Claude AI, Cowork and Claude Code will continue to exist as three separate products. Labs does two things, according to Krieger: “closing the gap” between what Claude can theoretically do and how it shows up in daily life, and acting as “advanced scouts” for where the models need to go next.
In practice that work means an “agent plus canvas” pattern for productivity software, with Claude Design as the first instance.
Anthropic launched Claude Design, described as agent-first mockups for non-designers. Long-horizon agents was the Labs project Krieger was leading, whose learnings became Managed Agents. A screenshot-verification step was demoed onstage by Boris.
Krieger is pushing on MCP as consumer distribution. A nonprofit search tool briefly surged to the top of Anthropic’s Connector list. Krieger views the episode as the early shape of how consumer AI breakouts will happen.
He pushed back on the idea that Anthropic is sandbagging the model for safety reasons, saying: “The internal belief is we should be able to release a Mythos-class model safely. ” Krieger abruptly stepped off Figma’s board after Anthropic launched Claude Design. ” Claude Design is a different use case from Figma’s more polished production and collaboration work, not a replacement.
The goal is not to be “the only product” in any space, which he said would be a tragedy. Health is the consumer white space that Krieger keeps returning to as an area where Anthropic should do more. 7 jump has finally made genetic and lab-data analysis with AI “actually useful versus cute” for the first time.
Anthropic’s January Function Health partnership was announced as the starting point for what could become a much bigger first-party health push. @alexeheath reported the podcast conversation in which Krieger laid out these views on product direction, organizational growth and future opportunities.
Single source — no framing comparison available.
globalnews.caTwenty-two member states pledged 30 to 35 gigawatts of new capacity by 2028 under the bloc's first tripartite deal. The European Commission will oversee annual progress tracking through 2028 as part of the Affordable Energy Plan.
zerohedge.comApple sued OpenAI and two former employees on July 10 in federal court in California. The complaint claims misappropriation of confidential engineering data and product details.
WiredFidji Simo will move to a part-time advisory position after extended medical leave. She joined OpenAI in May 2025 as CEO of Applications.