Anthropic has restructured Claude from a chat interface into a work environment: on November 10, 2026, the company opened three new surfaces to all paid tiers—Claude Whiteboards for planning, Claude Sheets for collaborative spreadsheets, and Claude Data for dashboards. The first two launched in beta, while the dashboards achieved general availability (GA) status and received a new name, replacing Claude Dashboards. The strategy is clear: instead of embedding AI into office applications, as Microsoft and Google do, Anthropic is embedding office applications into AI. Spreadsheets, boards, and dashboards live inside Claude, and the model directly sees their state and edits them.

image
image

What happened

On November 10, 2026, Anthropic announced the launch of Claude Whiteboards and Claude Sheets in beta status and simultaneously moved dashboards out of beta under the new name Claude Data, granting them GA status. All three products are available to subscribers of all paid tiers (Pro, Max, Ultra, Team, Enterprise) in web, desktop, and mobile applications; on the Enterprise tier, Claude Data will appear on November 19. Whiteboards is a shared board for brainstorming: a spoken plan is transformed into a structure with timelines and assignees, and the result can be exported to Lucid, Miro, and Vani by Zoho. Sheets is a spreadsheet that Claude creates and edits together with the user: formulas, tabs, and dropdown lists are supported, the finished file can be downloaded as .xlsx or CSV, and work can continue in Airtable, Asana, monday.com, Smartsheet, and Claude for Excel. Claude Data builds dashboards on top of corporate data: in beta, certified queries, data stories, and live sharing are available, with caching and versioning supported. Connectors work with Amazon Redshift, BigQuery, and ClickHouse, and partners include Amplitude, Mixpanel, and Omni. The video call assistant, Claude Meetings, has only been announced so far.

Context

Until now, Claude remained a dialogue: the model responded in chat, and users assembled finished documents in external tools. The new surfaces shift the interaction to a different format—work artifacts live inside the AI client, and the model gets persistent access to their state, editing the object directly rather than outputting text for copying. The market has still operated on the reverse scheme: Microsoft (Copilot) and Google embed AI inside their office suites, which already have an audience and contracts. Anthropic reversed the integration direction, leaving export bridges to Lucid, Miro, Airtable, and Excel as a safety path for transferring results. The company's own wording of the launch is telling: Sheets are described as spreadsheets with "simple functionality," meaning the authors do not promise full parity with mature spreadsheet products—full functionality is kept outside the model. From a scientific perspective, the announcement does not rely on new research: it is a product packaging of the model's existing agentic capabilities, not a new technical result.

Why this matters for the industry

For the industry, this is a direct attack on the territory of Microsoft 365 and Google Workspace with a different strategy: not "AI inside applications," but "applications inside AI." The blow lands on the Excel and Office combination, where Microsoft has long defended itself with its ecosystem and corporate contracts: if the artifact is born inside Claude, and third-party tools are only needed at the output, the usual entry point to a work document shifts. For product developers, there are two signals here. First—a new UX pattern: a live artifact to which the agent has persistent access changes the mechanics of interaction and raises the bar of expectations from "text generation" to "maintaining object state." Second—distribution: export bridges and interactive connectors turn other services into delivery channels for Anthropic's results. Startups in the workspace category now have to calibrate their pitches against model vendors, and protection shifts from the interface to data, vertical specialization, and distribution. However, of the three launches, only Claude Data passes the production test: Whiteboards and Sheets are betas with reduced functionality; their place is in pilots and rough work, not systems of record.

Why this matters for users

If you have a paid Claude tier, you can try all of this now without additional installations or switching between services. A practical scenario for today: plan a project on Whiteboards by speaking the task aloud and receiving a structure with timelines and assignees; assemble a working spreadsheet with formulas in Sheets and export it to .xlsx or CSV to finish in familiar Excel or continue in Airtable; connect Claude Data via connectors to your data warehouse and raise a dashboard. The zero cost of trial and export to familiar tools minimize the barrier. It is important to correctly choose the role of each product: Claude Data is the only one of the trio that has exited beta, and it should be perceived as an additional layer on top of your BI, not a replacement; Whiteboards and Sheets are suitable for drafts, sprint planning, and spreadsheet prototypes, but not for systems of record. Claude Meetings, the video call assistant, currently exists only in the announcement, so it cannot yet be integrated into the workflow.

What is still unknown / limitations

The announcement and its retellings do not contain a single benchmark, measurement methodology, or error metrics, and data on which specific models power the new features are not disclosed. Statements such as "Claude creates and edits a spreadsheet with formulas" and "certified queries" are presented without reproducible control, and functionality is currently confirmed only by the vendor's statement: the accuracy of formulas, the correctness of queries, and the behavior of dashboards on your data have not been independently verified. Parity with Excel has not been achieved or demonstrated, as indirectly indicated by the export to .xlsx and CSV for full spreadsheet functionality. Certified queries, data stories, and live sharing remain in beta; limitations on limits and costs are also not disclosed. Before implementation, it is reasonable to test the features on real data and track when the betas will move to GA and when Claude Meetings will be released.

Sources

Author

Look at AI, editorial team