Anthropic’s Cat Wu Says AI Will Soon Anticipate Users’ Needs Before They Know Them
Key Points
- Anthropic aims for a valuation near $950 billion in an upcoming funding round.
- Claude now leads among business customers, overtaking OpenAI and quadrupling market share since May 2025.
- Cat Wu, head of product for Claude Code and Cowork, emphasizes staying ahead of the AI curve rather than watching competitors.
- The firm released at least six new models last year and almost as many so far in 2026.
- Glasswing consortium—Amazon, Apple, CrowdStrike, Microsoft—gives partners early access to the cybersecurity model Mythos, which remains closed to the public for safety reasons.
- Wu envisions AI agents handling routine tasks, allowing workers to focus on creative work.
- Next AI milestone: proactive agents that anticipate user needs and automate workflows.
Anthropic’s product lead, Cat Wu, told reporters at the Code with Claude conference that the firm’s aggressive model rollout and focus on proactive AI will reshape work. With a funding round that could push Anthropic’s valuation near $950 billion, the company has already overtaken OpenAI among business customers, quadrupling its market share since May 2025. Wu highlighted the company’s relentless pace of model releases, the guarded launch of its cybersecurity model Mythos through the Glasswing consortium, and a vision of AI agents that handle routine tasks and anticipate user demands.
Anthropic is poised to eclipse its chief rival as it chases a valuation close to $950 billion in a funding round that could bring in tens of billions of dollars. The company’s rapid ascent is evident in its recent market data: business customers now favor Claude over ChatGPT, and Anthropic has quadrupled its share of enterprise users since May 2025, outpacing OpenAI.
At the second annual Code with Claude conference in San Francisco, Cat Wu, Anthropic’s head of product for Claude Code and Cowork, explained how the firm’s product strategy fuels that growth. Wu, who joined the company in August 2024, emphasized a relentless focus on staying at the "exponential" frontier of AI development rather than reacting to competitors. "If you spend time watching the competition, you end up two weeks or a month behind," she said.
The company’s output backs that philosophy. Anthropic released at least six new models last year and has already introduced nearly as many in 2026. Wu expressed confidence that the pace will continue, noting that each model iteration brings incremental improvements that can be shared quickly with users.
Safety remains a core concern. Earlier this year Anthropic launched Glasswing, a private consortium that includes Amazon, Apple, CrowdStrike and Microsoft, to give select partners access to Mythos, a powerful cybersecurity model designed to scan codebases for vulnerabilities. Unlike Anthropic’s other offerings, Mythos is not publicly released because the company fears the technology could be weaponized.
Looking ahead, Wu painted a picture of work transformed by AI agents. Managers will need to retain domain expertise to debug and guide these agents, but the ultimate goal is to shrink team size by automating repetitive tasks. "Everyone has a part of their job that’s tedious—like answering emails," she said. "We want Claude to handle that so people can focus on creative work."
The next milestone, according to Wu, is proactivity. While today’s AI assists users in real time, the upcoming generation will anticipate needs, set up automations, and reduce the friction of routine workflows. "Claude will understand what you work on and start setting up automations for you," Wu explained.
Anthropic’s trajectory, driven by an aggressive model release schedule, a guarded approach to high‑risk technology, and a vision of AI that works ahead of the user, positions it as a strong contender to lead the enterprise AI market in the coming years.