Anthropic brings the full Claude Desktop to major cloud platforms

Getting an enterprise IT team to approve a frontier AI model usually comes down to one question: where does our data actually go?

Anthropic answered that question clearly. The company expanded Claude Desktop across AWS, Google Cloud, and Microsoft Foundry, bringing the full suite—Chat, Claude Cowork, and Claude Code—into a unified desktop app.

Previously, enterprise users on these cloud platforms could only access Claude Cowork and Claude Code. Standard chatting was left out. Now, employees get all three surfaces under one roof, while enterprise admins retain total control over who sees what.

One app, three tailored surfaces

Rather than forcing every employee into the same workflow, the updated client segments roles neatly.

Non-technical teams can use standard Chat for quick answers or problem-solving. Claude Cowork handles labor-intensive research and document generation across approved internal sources and local files. Meanwhile, developers get Claude Code for agentic coding directly inside the app, freeing them from the command line.

Admins can manage each surface using separate policy keys. An organization can roll out Chat and Cowork to marketing and HR first, provision Claude Code exclusively for engineering, and expand access gradually. Hard-deny rules apply universally across every tab.

Built for strict IT guardrails

The real story here is the enterprise plumbing. Anthropic designed this deployment to slot directly into existing corporate infrastructure without headaches.

Inference stays entirely inside the customer's cloud environment within their chosen regions, while conversation histories reside locally on the device. Telemetry shared back to Anthropic remains under admin control.

Authentication integrates with standard SSO providers, including AWS IAM Identity Center, Google Cloud Workforce Identity Federation, Microsoft Entra ID, and Okta. End users never touch raw cloud credentials or shared API keys. For deployment, IT teams can push MDM policy templates via Microsoft Intune, GPO, or Jamf, and an offline installer supports air-gapped environments.

For teams working inside Microsoft 365, a dedicated connector pulls emails and files via custom Entra apps. Stricter organizations can run a local connector that keeps traffic strictly between the endpoint and Microsoft, with beta support for GCC High and DoD environments.

Why it matters

We think this is one of the smartest enterprise distribution plays Anthropic has made to date.

Most enterprise AI adoption stalls because security officers refuse to let corporate data touch public web endpoints, while developers hate clunky internal portals. By packaging Chat, Cowork, and Code into a single desktop native experience that routes all inference through an existing AWS, Google Cloud, or Microsoft bill, Anthropic eliminates almost every legal and architectural hurdle at once.

It also solidifies Anthropic’s position as cloud-agnostic infrastructure. Instead of forcing companies into a proprietary platform silo, it lets businesses run their AI models inside the sovereign cloud envelopes they already pay for and trust.

The trade-off is management overhead: IT administrators now have to configure policy templates, validate connectors, and map identity roles before users can touch it. But for companies already invested in the big three clouds, this is easily one of the cleanest and safest ways to roll out agentic AI across an entire workforce today.