Claude Cowork, available on macOS, is losing the “research preview” label today as Anthropic introduces enterprise capabilities. Separately, Anthropic is reclaiming the AI agent narrative with Claude Managed Agents.

Claude Cowork is ready for enterprise

Anthropic introduced Claude Cowork three months ago as a research preview. Since then, the ability to let Claude manage workflows on the Mac has only matured.

Starting today, Claude Cowork includes six features built especially for enterprise use:

• Role-based access controls
• Group spend limits
• Usage analytics
• Expanded OpenTelemetry support
• Zoom MCP connector
• Per tool connector controls

You can learn more about each feature in the company’s announcement post.

Anthropic says Claude Cowork for macOS and Windows is now generally available for all paid subscribers.

Claude Managed Agents are here

Meanwhile, Anthropic has a new public beta tool called Claude Managed Agents. Managed Agents are “a suite of composable APIs for building and deploying cloud-hosted agents at scale,” the company says.

Until now, building agents meant spending development cycles on secure infrastructure, state management, permissioning, and reworking your agent loops for every model upgrade. Managed Agents pairs an agent harness tuned for performance with production infrastructure to go from prototype to launch in days rather than months.

Companies like Notion, Asana, and Sentry have already used Managed Agents to create new solutions with Claude. Now the new tool is available on the Claude Platform for all builders.

Both developments follow the introduction of Claude Mythos Preview and Project Glasswing. Anthropic isn’t releasing its most powerful AI model that it announced this week.

Instead, it’s working with companies, including Apple, to use Mythos as a cybersecurity tool for discovering and resolving software vulnerabilities.