← Back to Guides

Claude Work Reference

A practical reference for Claude Cowork and Claude for Work — agentic task execution, permission modes, projects, scheduling, and enterprise admin controls.

Last updated 2026-09-03
Claude
Claude Cowork
Claude for Work
Enterprise
Admin

Claude Work Reference

Claude Cowork, Claude for Work, and Claude Enterprise

2026 Edition


About this guide

"Claude Work" spans a few related but distinct things, and it's worth being precise about which is which:

  • Claude Cowork — an agentic interface that lets Claude execute multi-step tasks on your behalf, in the cloud, with configurable approval workflows. Built on the same agentic architecture that powers Claude Code, but aimed at knowledge work broadly rather than coding specifically.
  • Claude for Work / Claude Enterprise — Anthropic's plan tier and product bundle for organizations: Claude Chat, Claude Cowork, and Claude Code together, plus admin, security, and compliance tooling (SSO, RBAC, audit logs, spend controls).
  • Claude Team — the smaller-organization tier; Cowork is included on both Team and Enterprise seats.

This guide covers Cowork's features and workflow, plus the admin/security surface of Claude for Work and Claude Enterprise. It does not cover the consumer Claude.ai chat experience (Projects, Artifacts, memory, styles) or Claude Code — see the separate guides for those.

Sources are Anthropic's own Help Center (support.claude.com) and product pages (claude.com/solutions/enterprise).


Table of contents

  1. What Claude Cowork is
  2. Availability by platform
  3. Getting started with Cowork
  4. Permission modes
  5. Projects in Cowork
  6. Scheduled tasks
  7. Usage considerations
  8. Claude for Work / Claude Enterprise
  9. Admin and security controls
  10. Official resources

1. What Claude Cowork is

Claude Cowork enables Claude to take on complex, multi-step tasks and execute them on your behalf — you describe an outcome, step away, and come back to finished work, rather than staying in a synchronous back-and-forth chat.

Key distinction from regular chat: standard Claude.ai chat is conversational — Claude responds, you respond, and nothing happens without you present. Cowork sessions run in the cloud and continue even when you close your laptop or step away, with Claude analyzing the request, creating a plan, breaking complex work into subtasks, and coordinating parallel workstreams when needed.

What Cowork can do:

  • Long-running tasks without conversation timeouts.
  • Local file integration — on desktop, Claude reads and writes files directly, without manual uploads or downloads.
  • Professional outputs — spreadsheets with formulas, presentations, and formatted documents, not just raw text. Outputs can be further edited with Claude for Excel and Claude for PowerPoint.
  • Web and browser use — for tasks that touch websites, Claude can open pages, read content, click, type, and fill forms. On desktop, this uses a browser built into the Claude Desktop app by default.
  • Scheduled and recurring tasks — see Scheduled tasks below.

2. Availability by platform

Platform Availability
Claude Desktop (macOS/Windows) All paid plans
Web (claude.ai) Pro, Max, Team; Enterprise where admin-enabled
Mobile (iOS/Android) Pro, Max, Team; Enterprise where admin-enabled
Chrome side panel Rolling out to paid plans; Enterprise where admin-enabled

Cross-device continuity: sessions follow your Claude account across devices — start a task on desktop, monitor it from mobile, and retrieve the finished output from web.

Local access requirement: keep the Claude Desktop app open when a task needs local file access, browser use, or computer interaction — even though the task itself runs in the cloud.


3. Getting started with Cowork

  1. Open Claude on any supported surface.
  2. Select Cowork in the message box.
  3. Describe your task.
  4. Review Claude's approach and let it execute.

For a guided setup on desktop, Anthropic's documented three-step flow is: open the desktop app, run the guided setup, and point Cowork at where the work lives. In your first session, typing /setup-cowork walks you through selecting your role, installing a matched plugin, and connecting tools like Slack, Google Workspace, or Microsoft 365.


4. Permission modes

Three approval modes control when Claude asks before taking an action in Cowork:

Mode Behavior
Manual Pauses for your explicit approval on each action
Auto Automatically approves actions it judges safe, blocks anything risky
Skip No automatic checking — use only for fully trusted tasks

Connector tool approval settings can also vary by role within an organization, and full automatic-approval mode may require admin enablement in Team/Enterprise settings.


5. Projects in Cowork

Projects in Cowork group related tasks together with their own files, links, instructions, and memory — the same underlying concept as Projects in Claude.ai chat, but oriented around ongoing work rather than conversation history. Projects are available to all Cowork users; there are currently no separate admin controls to restrict who can create a project at the organization level.


6. Scheduled tasks

Cowork supports recurring tasks that run automatically on a schedule, or on demand, independent of whether your device is online — since the work itself executes in the cloud. This is useful for things like a daily status digest, a weekly report, or routine cleanup work that shouldn't require you to remember to kick it off manually.


7. Usage considerations

Cowork tasks are compute-intensive: complex, multi-step work consumes significantly more of your usage allocation than a standard chat turn, since it requires more tokens to plan and execute autonomously.

Guidance: reserve Cowork for genuinely complex, multi-step work that benefits from autonomous execution. Use standard chat for simpler, single-turn questions.

Security note: agentic interfaces carry a non-zero risk of prompt injection — content Claude reads (a web page, a file, an email) could contain instructions intended to manipulate its behavior. Avoid sharing highly sensitive files with agentic sessions, and monitor Claude's actions, especially early on while you're calibrating trust in a given workflow.


8. Claude for Work / Claude Enterprise

Claude for Work bundles the core products for organizational use:

  • Claude Chat — the everyday conversational assistant for drafting, research, and working through problems.
  • Claude Cowork — delegated, multi-step task execution (see above); connects to business applications and hands off polished deliverables.
  • Claude Code — coding across the terminal, IDE, Slack, or web, for natural-language-driven development work.
  • Claude Security — code review for security issues, vulnerability explanations, and patch drafting.

Claude Enterprise is the top tier of this bundle, adding an expanded context window (up to 500K tokens), higher usage capacity, and native GitHub integration for working across entire codebases.


9. Admin and security controls

Identity and access:

  • Single sign-on (SSO/SAML)
  • Domain capture
  • SCIM provisioning
  • Role-based access control (RBAC), including user groups with custom roles

Compliance and data protection:

  • SOC 2, ISO 27001, GDPR, and CCPA certified; HIPAA-ready offering
  • No model training on customer data by default
  • Data retention controls (Enterprise)

Monitoring and governance:

  • Audit logs and OpenTelemetry monitoring (Enterprise)
  • Usage analytics and reporting, including an admin analytics dashboard showing usage and cost by group and by user, filterable by the SCIM groups your organization already manages
  • Per-user spend caps and spend-threshold alerts
  • Model-level entitlements — admins can control which models are available to which groups
  • A Compliance API for Enterprise

Connector governance: organizations can control which connectors are available and set per-connector-tool approval requirements (e.g. requiring explicit approval for certain actions rather than allowing automatic approval).


10. Official resources

Topic URL
Get started with Claude Cowork https://support.claude.com/en/articles/13345190-get-started-with-claude-cowork
Use Claude Cowork on Team and Enterprise plans https://support.claude.com/en/articles/13455879-use-claude-cowork-on-team-and-enterprise-plans
Use Claude Cowork on web, desktop, and mobile https://support.claude.com/en/articles/15520349-use-claude-cowork-on-web-desktop-and-mobile
Organize tasks with projects in Claude Cowork https://support.claude.com/en/articles/14116274-organize-your-tasks-with-projects-in-claude-cowork
Assign tasks from anywhere in Claude Cowork https://support.claude.com/en/articles/13947068-assign-tasks-from-anywhere-in-claude-cowork
Claude Cowork collection https://support.claude.com/en/collections/19667525-claude-cowork
Claude Enterprise https://claude.com/solutions/enterprise
Claude for Enterprise (product overview) https://claude.com/blog/claude-for-enterprise
Anthropic Help Center https://support.claude.com/en

Final accuracy note

This guide reflects Anthropic's own Help Center and product-page documentation as reviewed on September 3, 2026. Enterprise and admin features in particular change frequently as Anthropic ships new governance tooling; confirm current details in the Anthropic Help Center or with your organization's Claude admin before making decisions that depend on a specific control being available.


Claude Work Reference — 2026 Edition

Put this into practice with Prompt Architect

Turn what you just read into production-ready prompts and workflows — free for 7 days, no credit card required.

Start free trial