Anthropic has launched interactive applications inside its Claude AI assistant, allowing users to connect and operate popular workplace tools without leaving the chat window.

The initial rollout includes integrations with Asana, Figma, Canva, Slack, Box, Amplitude, Clay, Hex, and Monday, with Salesforce support planned for a future update. These apps allow Claude to read, write, and act across tools while keeping the user in a single conversational interface.

The integrations are built on MCP Apps, an open extension of Anthropic’s Model Context Protocol. This design allows other AI platforms to implement similar functionality, positioning MCP as a potential cross-platform standard for AI-to-app interaction.

All actions require explicit user consent before execution, and enterprise administrators can restrict which tools employees are allowed to access. Anthropic emphasized these controls as essential for enterprise safety and governance.

The feature is available at no additional cost to Pro, Max, Team, and Enterprise subscribers. Anthropic also confirmed that a future integration with its Cowork product is planned.

Why it matters: Frontier AI labs are racing to become the primary interface layer for work. As context becomes the key advantage in AI usefulness, assistants that sit atop existing productivity tools may increasingly replace traditional app-switching workflows.