- What did Anthropic ship on November 10?
- How far does Claude Sheets go today?
- What changes when the whiteboard keeps the plan?
- Why is Claude Data the bigger decision?
- What happens on November 19?
- What should a Canadian team test this week?
Anthropic shipped 3 office tools on November 10, 2026, and the one that looks like a spreadsheet is the least important of the 3. Claude Sheets and Claude Whiteboards are now in beta on paid plans, and Claude Data, a live view of the files, people and work Claude can reach, is already switched on. The sheet is a clean surface for a calculation. The data layer decides what Claude may touch, and on Enterprise plans it switches itself on in 9 days unless an admin stops it.
The direct answer: Claude Sheets is a lightweight workbook for turning a chat into a table, a chart or a small model. Claude Data is the access layer behind it, and on Enterprise plans it turns on by itself on November 19, 2026. A paid user can open Sheets and Whiteboards in beta today. An Enterprise admin who does not want Claude reading the workspace should open the admin controls before that date.
What did Anthropic ship on November 10?
Anthropic shipped 3 products that look like one office suite. Claude Sheets is a workbook that builds tables, charts and models from a prompt, with a formula bar that accepts plain language instead of syntax. Claude Whiteboards is a canvas where a team plans work and Claude turns the plan into a working artifact, including an app or a deck. Claude Data is the connective layer: a live index of files, people and work across a connected Google Workspace, Microsoft 365, Slack, Asana, Notion and Salesforce.
Anthropic also opened its interactive connectors to the public beta, so third-party apps can surface inside Claude the same way its own tools do. The announcement treats the 3 products as one bet, that the model is most useful when it can see the work. On that point, the bet matches a pattern this magazine already tested in Notion agent Memories are private to each user. Decide what they may keep: the value of a connected assistant sits in what it remembers, and the risk sits in the same place.
How far does Claude Sheets go today?
Claude Sheets is useful and deliberately small. A spreadsheet that takes instructions in plain language will still hand you a broken model if you ask it to plan a year of hiring. The product is in beta, and the limits are visible in what Anthropic says it does well: quick tables, a chart, a small model, a formula explained in words. It is not a billing system and it is not a ledger. Treat it as the model doing the typing while you keep the arithmetic.
Anthropic states the limit in its own launch post. The pattern also showed up in the plan Claude sells as unlimited: a judge found that “20x” was not a promise the contract supported, a point this magazine covered in A judge says ’20x’ has to mean 20x. Read your AI plan like a contract. Claude Sheets is best read the same way, as a tool with a stated scope, not a promise.
What changes when the whiteboard keeps the plan?
Whiteboards are the more interesting product because they keep the plan, not just the artifact. A whiteboard that remembers a decision can turn a Tuesday sketch into a Thursday deck without anyone retyping the brief. That is also the risk. A plan that stays in the room keeps its assumptions, including the ones nobody checked. Claude Whiteboards is in beta on paid plans, and the product page is the right place to watch for what it retains. The useful test is simple: write a plan on a whiteboard, let Claude build the artifact, then ask it a week later what the plan assumed. If it can answer, the whiteboard is a memory. If it cannot, it is a sketchpad.
Why is Claude Data the bigger decision?
Claude Data is the decision that outlives the other 2 products because it changes what Claude can reach, not just what it can draw. A workbook and a whiteboard are surfaces. Data is the wiring behind them: a live view of files, people and work across Google Workspace, Microsoft 365, Slack, Asana, Notion and Salesforce. A live index is a convenience for a team that trusts its house, and a liability for a team that does not. The point is not that Claude reads your files. The point is that someone decides which files it may read, and when that decision changes. On Enterprise plans, the answer is that the decision changes on November 19, 2026, unless an admin stops it.
What happens on November 19?
On November 19, 2026, Claude Data will turn on by itself for Enterprise plans that have not opted out. The wording matters: this is an opt-out, not an opt-in, and the window to act is the 9 days between the launch and the switch. An Enterprise admin who misses the date does not lose control, but does lose the quiet version of it: Claude gains access to the connected workspace until someone turns it off. The practical move is small and specific. Open the admin console, find the Claude Data setting, and decide before November 19 whether the workspace is in or out.
The same caution applies to the smallest habits, the ones that look harmless until they are not: a model that can read your files is only as safe as the file you pasted into it. This magazine has covered the file-paste risk in detail, first in Claude Personal Finance Canada is still US-only. Ban bank CSV pastes first and then in One Claude Code setting leaked 1,379 passwords and keys. The check takes 10 minutes. The file you paste is the file the model keeps.
What should a Canadian team test this week?
Test the sheet before you test the wiring. Open Claude Sheets, hand it a real table from your own work, and ask it to build a chart and explain one formula in words. If the explanation holds, the tool saves time. Then open the admin console and look at Claude Data before November 19. If the workspace is connected and the admin has not decided, decide. The sheet is the demo. The data layer is the decision. Test the first this week, and make the second before the 19th.
The recap is short. Claude Sheets is a clean workbook for a quick model, Claude Whiteboards keeps the plan, and Claude Data decides what the model may reach. The sheet is the least of it. The data layer is the real call, and on Enterprise plans the call makes itself on November 19 unless an admin steps in.
If your firm runs Claude on an Enterprise plan, open the admin controls today and decide whether Claude Data is in or out before the 19th.
