Anthropic is shifting gears, moving from chatbots to agentic AI that can perform real tasks. The AI startup is also dialing up focus on trusted, enterprise-ready systems.
The company raised $3.5 billion in March, pegging its valuation at $61.5 billion. They launched a blog dedicated to their Claude models and reportedly teamed up with Apple on a "vibe-coding" platform.
At this week’s TechCrunch Sessions: AI in Berkeley, co-founder and CSO Jared Kaplan talked to Equity co-host Max Zeff. 
Kaplan revealed who gets direct access to Claude’s models and how Anthropic balances openness, safety, and sustainability. He emphasized the company’s pivot beyond chatbots toward AI agents that do actual work, highlighting internal tools like Claude Code shaping AI-powered development.
Kaplan also touched on building AI enterprises can trust and the human impact on work and software interaction.
Listen to the full episode on Apple Podcasts, Spotify, or Overcast.
Jared Kaplan stated:
“Who has direct access to Claude’s AI models, Windsurf’s response, and how it all ties into Anthropic’s broader goals around openness, safety, and sustainability.”
“The company’s pivot away from chatbots and toward agentic AI systems that can perform real tasks.”
“How internal tools like Claude Code are shaping the future of AI-powered development.”
“What it means to build AI that enterprises can actually trust, and how that affects the way humans interact with software, work, and each other.”
