Claude Cowork and Chat Merge on Enterprise, With Spend Controls
Anthropic made its unified Cowork and chat experience generally available on Claude Enterprise plans on October 23, 2026, ending the split between asking Claude a question and handing it a project. A conversation starts as chat; when a task needs code or file work, Claude continues in the cloud and returns the finished document, analysis or deck. Work keeps running after the laptop closes and resumes on web, desktop or mobile, with chats, projects, connectors, skills and plugins carried over, per Anthropic’s announcement.
The governance details matter more than the demo. SSO and the Compliance API still apply, data is not used for training by default, and each agentic session runs in its own isolated virtual machine with network egress off by default for Enterprise. Pricing stays usage-based. Anthropic says Claude Opus 5.5 costs 40% less to run than Opus 5 on typical workloads, and Sonnet 5.5 up to 30% cheaper than Sonnet 5. Admins get four cost controls: efficient mode, model entitlements by role, spend limits and pooled group budgets. Those extend the per-group cost breakdowns and spend alerts Anthropic added for Enterprise admins in July.
The path has been visible since August, when the Claude in Chrome side panel became a full Cowork session that carries conversations, skills and connectors across apps. Enterprise plans had that feature off by default, with administrators able to switch it on.
The outcome question is who owns review. LifeMoves said Claude produced a report in under ten minutes that would have taken its team weeks — a customer claim, not an independent measurement. Delegating whole projects ships results only when someone signs off on the output. Before rollout, decide which project-sized tasks go to the agent, who approves what comes back, and which spend limit applies. Organizations on zero data retention, a HIPAA-ready configuration or customer-managed keys stay on their current setup, so check that first.