Anthropic Turns Claude Into a Full Office Suite With Sheets and Whiteboards

Anthropic is rolling out three new Claude work surfaces: Claude Sheets, Claude Whiteboards, and Claude Data. They let users build spreadsheets, organize ideas on a canvas, and generate dashboards without leaving a conversation. Upgraded connectors can also embed apps from companies including Adobe and Atlassian inside Claude.

The design extends the interaction model Anthropic established with Claude Code. Users describe an outcome, inspect the result, and revise it with Claude. The new tools apply that loop to spreadsheets, documents, dashboards, and other knowledge-work artifacts.

Three artifacts share one context

Each artifact acts as an editable document that people and Claude can develop together:

• Claude Sheets creates and edits spreadsheets, formulas, data structures, and multiple tabs.
• Claude Whiteboards provides a canvas for arranging ideas during a meeting and can summarize the resulting board.
• Claude Data, previously called Dashboards, connects to systems such as Snowflake and Salesforce. It can use definitions certified by a company’s data team to produce dashboards, visual data stories, and narrated short videos.

Claude carries conversational context between these artifacts, allowing a team to move an idea from a whiteboard into a spreadsheet and then a dashboard without repeatedly supplying the same background. Users can also request custom artifacts such as account trackers, project status boards, and contract review tables.

Connectors bring apps into Claude

Interactive connectors allow partners to embed their application interfaces directly in Claude. Users can edit content themselves inside the embedded app or ask Claude to make changes alongside them, reducing the copying and tab switching required by earlier connector workflows.

Plugins from Salesforce, currently in beta, NetSuite, and Workday, also in beta, include task-specific skills. Anthropic lists sales-call preparation, financial analysis, time-off requests, and expense handling among the supported workflows.

These integrations use the Model Context Protocol, Anthropic’s open standard for connecting AI systems to external tools and data. Claude Marketplace now lists more than 9,000 connectors and plugins, including community-built integrations.

Who gets access

Accounts will receive the features in stages. Anthropic says access follows its existing paid tiers, with no separate product SKU. The forthcoming Claude Meetings feature will join calls, take notes, and update documents during the conversation.

Early users report shorter cycles

Anthropic published several internal and customer examples, though the company did not provide independent benchmarks or detailed testing methods:

• Anthropic’s legal team built a Slack-based review tool that generates redlines from its playbook, reducing marketing review time from three days to 24 hours.
• Advantage Solutions’ financial planning and analysis team reduced the preparation of forecast slides, commentary, and validation from about 10 hours per cycle to less than 30 minutes.
• HubSpot’s customer success team uses live dashboards connected to HubSpot and drafts follow-up emails within 30 minutes of a call ending.

Enterprise controls narrow access

Putting company data, spreadsheets, and business applications in one AI interface raises questions about permissions, credentials, and spending. Enterprise administrators can choose which employees may use specific products and connectors, set granular organizational spending limits, and restrict the actions available through each integration.

Claude follows the permissions already assigned to a user. Administrators can further limit connector capabilities, allowing Claude to read Slack messages, for example, while withholding permission to send them.

Anthropic is also testing organization-level setup through its auto-connect setup guide. The feature lets administrators configure supported connectors once for the organization. In the new Enterprise experience, every cloud session runs in an isolated virtual machine, while connector credentials remain outside that environment.

Anthropic wants to own the work surface

Anthropic’s product strategy centers Claude as the place where work is created and revised. Earlier workflows often required users to paste context into a chat, generate an answer, and transfer the result into Excel, Notion, Figma, or another application. Artifacts and embedded connectors keep more of that process inside one interface while preserving the conversation’s context.

The expansion puts Claude in closer competition with AI features in Google Workspace, Microsoft 365, and Notion. Anthropic’s pitch is a shared context layer that spans multiple artifacts and third-party tools, giving teams a single starting point for trackers, analyses, dashboards, and other operational work.

The release packages existing model behavior and MCP integrations into additional work surfaces. Anthropic’s announcement leaves several practical questions unanswered, including formula reliability in large spreadsheets, whiteboard summarization during messy brainstorming sessions, connector latency, and the fidelity of embedded apps compared with their native interfaces. Evaluations should use representative data, permission settings, and multi-step workflows before these tools support financial forecasts or other consequential processes.