Claude Workspace Guide (2026): Console, Cowork, Google

Last Updated: November 11, 2026

Summary: A Claude workspace is, officially, a container inside the Claude Console that groups API keys, members, spend limits and rate limits so a company can separate projects, teams or environments under one Anthropic organization. Every organization gets a Default Workspace and can create up to 100 by default. People also use "Claude workspace" for project spaces in Claude Cowork and for Claude for Google Workspace, the add-on that puts Claude inside Docs, Sheets and Slides.

The phrase is confusing because Anthropic uses the word in several places. A developer setting spend caps and an IT admin rolling Claude out to Google Docs can search the same two words and mean different things.

This guide covers every official meaning, starting with Console workspaces, the feature Anthropic actually names "Workspaces." Features, limits and prices were checked against Anthropic's documentation, help center and pricing page on November 11, 2026.

A Claude workspace most often means a Workspace in the Claude Console: a unit that organizes API usage inside one organization while billing and administration stay central. According to Anthropic's Workspaces documentation, teams use workspaces to separate projects, environments or teams.

The term shows up in three other places in Anthropic's own materials, so it helps to know which one you need before you start clicking.

If you manage API keys or an API budget, read the Console sections. Subscription users can skip to the Google Workspace, Cowork and admin sections.

Console workspaces are separate environments inside one Anthropic API organization, each with its own API keys, members, usage reporting and limits. Anthropic introduced them on September 10, 2024, describing them in its launch announcement as a way to manage multiple Claude deployments, such as development, staging and production.

The basic rules for Claude Console workspaces, from Anthropic's documentation:

Claude Console workspaces have five roles, and organization users and developers must be added to each workspace explicitly. Organization admins automatically become Workspace Admin on every workspace, and billing members automatically get Workspace Billing.

Each Claude Console workspace can have its own monthly spend limit and rate limits, but only lower than the organization's limits, never higher. If a workspace has no limits set, it inherits the organization's limits, and organization-wide limits always apply even if workspace limits add up to more.

Rate limits can be set per model. The Claude help center adds that admins can set email notifications when a workspace's spend reaches a chosen amount, and can view usage and cost reports per workspace or across all of them.

A Claude Console workspace is also a data boundary: every API request runs in exactly one workspace and can only reach resources in it, including files, message batches and skills, according to Anthropic's documentation.

Keys work in two ways. A key scoped to a single workspace always runs there. A personal or service account key that is not scoped to a workspace runs in whichever workspace the request names in the anthropic-workspace-id header, as long as the account has access. Prompt caches are isolated per workspace on the Claude API, Claude Platform on AWS and Microsoft Foundry, but per organization on Amazon Bedrock and Google-managed Claude on Google Cloud.

An organization admin creates a Claude Console workspace from Settings, then Workspaces, in about a minute. The steps from Anthropic's documentation and help center:

Archiving deserves care. Archiving a Claude Console workspace keeps its historical data for reporting, but it deactivates the workspace, archives every API key created for it within seconds, and cannot be undone.

The Claude Code workspace is a special Console workspace that Anthropic creates automatically the first time someone in an organization signs in to Claude Code with a Console account. Anthropic's Claude Code cost documentation says it provides centralized cost tracking for all Claude Code usage in the organization, and that you cannot create API keys for it.

The Claude Code workspace has three quirks worth knowing:

On Claude Team and Enterprise subscriptions, Claude Code usage draws from each member's seat allowance instead, managed in claude.ai admin settings. Our guide to Claude usage limits explains how those allowances work, and our Claude Code review covers the tool itself.

Claude workspaces also exist on Amazon's cloud, where they double as AWS IAM resources. On Claude Platform on AWS, every inference request targets a workspace through the anthropic-workspace-id header, and each workspace is bound to a single AWS region, according to the AWS user guide for Claude Platform.

Amazon Bedrock has a separate feature called Workspaces (Anthropic-compatible) for the Anthropic Messages API on its bedrock-mantle endpoint. The Amazon Bedrock documentation describes them as boundaries for access control, cost tracking and observability.

The simplest workspace setup for most companies is one workspace per environment, plus separate workspaces for any team or customer whose spend you need to track on its own. Anthropic's launch post uses development, staging and production as the core example.

Here is a hypothetical example to show how the limits interact. The dollar figures are illustrative, not Anthropic recommendations:

Suppose the organization's own limit is $7,000 a month. The three workspace caps add up to $8,000, but Anthropic enforces the organization limit on every request, so total spend still stops at $7,000. A runaway test script in development hits its $500 cap long before it can eat into production's budget, which is the main reason to separate environments at all.

Two habits follow from Anthropic's rules. Create production keys inside the production workspace, because keys cannot be moved later. And keep real traffic out of the Default Workspace, since it cannot have a spend limit.

No. Claude for Google Workspace is a separate product: an add-on that opens Claude in a sidebar inside Google Docs, Sheets and Slides. Anthropic announced on October 6, 2026 that the add-on is "now in public beta on all paid Claude plans," in its launch article.

The Claude help center lists the add-on as available in beta for Pro, Max, Ultra, Team and Enterprise plans. Users install it from the Google Workspace Marketplace, then open a file and choose Extensions, then Claude, then Open Claude. One install covers all three apps.

What Claude for Google Workspace can do, according to the help center:

By default the add-on asks before each edit, and Claude cannot share the file or change access. Google Workspace super admins can install it for the whole organization or selected groups. The help center says the add-on does not support Zero Data Retention or HIPAA coverage.

Separately, Claude's Gmail, Google Calendar and Google Drive connectors let Claude search and act on those services from inside a Claude chat. The help center article on Google Workspace connectors says they are available for all users on Claude and Claude Desktop. On Team and Enterprise plans, an Owner or Primary Owner must enable them first, and Claude asks for approval before sending email or sharing files by default.

Connectors are built on the Model Context Protocol, which our explainer on the Claude MCP server covers in detail.

In Claude Cowork, Anthropic's help center says projects let you group related tasks into "dedicated workspaces," each with its own files, context, instructions and memory. The Cowork projects help article says a project can include a local folder, a linked chat project or a pasted URL, plus recurring scheduled tasks.

Cowork projects are available on Claude Desktop for all paid plans: Pro, Max, Ultra, Team and Enterprise. Web and mobile access covers Pro, Max, Ultra and Team (in beta on Team), plus Enterprise when an owner enables it. Sharing projects is limited to Team and Enterprise plans.

Cowork can read and write local files only in folders you connect, according to Anthropic's getting started guide. Our article on Claude computer use covers the related ability to operate desktop apps, and our guide to Claude AI agents explains how Claude handles multi-step work on its own.

On Claude Team and Enterprise plans, the shared space is called an organization, not a workspace, and admins manage it from Organization settings on claude.ai. Owners control connectors, Cowork, capabilities and spend limits there.

Key admin controls from Anthropic's help center, checked November 11, 2026:

Prices come from Anthropic's pricing page and exclude tax. Console workspaces are not tied to these plans: they are part of the pay-as-you-go Claude API.

The right Claude workspace depends on whether you build with the API, organize your own work or roll Claude out to colleagues.

Many businesses will use more than one. For a wider view of what Claude does, see our complete guide to Claude AI.

Sameer's take: Claude is my favourite of all the AI tools.

What is a workspace in Claude?
In Anthropic's terms, a workspace is a container in the Claude Console that groups API keys, members, usage reports, spend limits and rate limits inside one organization. Claude Cowork also describes its projects as workspaces with their own files, instructions and memory.

How many workspaces can a Claude organization have?
A Claude Console organization can have up to 100 workspaces by default, according to Anthropic's documentation. Archived workspaces do not count toward the limit, and Anthropic says to contact your account team if you need more.

Can I set a spend limit on a Claude workspace?
Yes. Admins can set a monthly spend limit and spend notifications on any Console workspace except the Default Workspace, and the limit must be lower than the organization's limit. The organization limit always applies on top.

Is Claude for Google Workspace free?
Claude for Google Workspace is in beta for paid Claude plans: Pro, Max, Ultra, Team and Enterprise. Anthropic has not published a separate price for the add-on, so it runs on your existing Claude plan.

What happens when you archive a Claude workspace?
Archiving a Console workspace keeps its historical data but deactivates it and archives all of its API keys within seconds. The action cannot be undone, and archiving the Claude Code workspace blocks Claude Code sign-in through Console billing for the whole organization.

A Claude workspace, in Anthropic's own vocabulary, is a Console container for API keys, members and limits, and every API organization should use them to separate production from testing. The Claude Code workspace, AWS workspaces, Cowork projects and Claude for Google Workspace share the name but solve different problems.

Developers need Console workspaces, knowledge workers need Cowork projects, and Google-first companies should look at the add-on, still in beta as of November 2026.

⭐ Get more AI Business Weekly in Google: add us as a preferred source on Google.

Written by Sameer Khan with AI assistance, fact-checked against the sources linked above.