A workbook is a shared analysis context for selected data. It holds source structure, table relationships, questions, and dashboards. Source permissions and freshness determine what each question can use.
A Workbook, defined
Think of a Workbook as a container, not a file. A single spreadsheet tab holds one flat table. A CSV upload holds one flat table, frozen the moment you exported it. A Workbook holds as many sources as an analysis actually needs — a renewals sheet in Google Sheets, a subscriptions table in MySQL, a support-tickets collection in MongoDB — treated as one connected whole instead of several separate files you reconcile by hand.
Return to the same workbook for a recurring business review. Keep useful questions and dashboards together, and check source status before comparing results from different dates.
That's also the direct contrast with a single flat sheet. A flat sheet is one table, one shape, done as soon as you save it — fine for a single answer, but it can't hold a second source, and it has no way of knowing when the numbers behind it have moved on. A Workbook is built to hold more than one source and to notice when any of them changes.
What you can put in it
A Workbook isn't limited to one kind of source, and it isn't limited to just one of each either — add more than one sheet or more than one database if the analysis calls for it. You can connect:
Google Sheets and Excel bring the spreadsheets your team already works in. MySQL and MongoDB bring the databases your product runs on, queried in place instead of exported first, and more connectors are being added regularly. If your data lives in PostgreSQL or Druid instead, you can already query those directly in plain English — that's a closely related capability covered on the talk to your database page.
How sync works
Freshness depends on the connector. Database questions can query the source on demand. Google Sheets, cloud billing, and advertising connections use refreshed snapshots. Excel, CSV, and Tally imports reflect the supplied data until updated.
Workbook
Synced
Check the source and reporting period behind an answer. An up-to-date database query and an older spreadsheet snapshot can both be valid individually while producing a misleading combined comparison. Align the periods and refresh state first.
Asking across sources
The payoff of putting multiple sources in one Workbook is that a single question can use all of them. Instead of pulling one number from a sheet, another from a database, and reconciling them by hand, you ask in plain English and tableArth.ai pulls from every connected source to return one unified answer.
You only ask once. tableArth.ai works out which sources and tables the question actually needs, joins them behind the scenes, and returns a single answer — with the chart that fits it — instead of three separate results you'd have to combine yourself. Nobody asking the question has to know which source held which number; that's the join tableArth.ai makes for you.
Why it beats a flat export
Compare that to the default most teams fall back on: exporting everything into one flat sheet. It works, right up until the data changes or the analysis needs more than one table.
| Capability | One-time export / flat sheet | Workbook in tableArth.ai |
|---|---|---|
| Freshness | Frozen the moment you exported it | Live — matches each source right now |
| Structure | Tabs and tables squashed into one | Each source's tables stay intact |
| Staying current | Re-export and re-upload by hand | Syncs on its own, nothing to re-upload |
| Sources in play | One file, one source | Many sources, connected side by side |
| Cross-source questions | Not possible — nothing to join across | One question, answered across every source |
A flat export also throws away structure. Multiple tabs or tables get squashed into a single sheet, which means VLOOKUPs and copy-paste before you can even ask a question. A Workbook keeps each source's tables and tabs intact and joins related data the way you'd join tables in a database, so your data never has to be flattened into one sheet just to be usable. See how this plays out day to day in live data vs. one-time uploads, or how the joins themselves work in joining data across sheets and tables.
The throughline is the same one running through the rest of this page: a Workbook is data you keep connected, not data you keep re-sending. Build it once, and every question you ask after that is answered against what's actually true right now.
What is a Workbook?
A TableArth workbook groups selected sources, their tables, questions, and dashboards so people can analyze business data in a shared context. It preserves the access and refresh behavior of each source.
What sources can it hold?
A Workbook can hold Google Sheets, Microsoft Excel files, and databases such as MySQL and MongoDB, side by side in one place — with more connectors on the way.
Does it stay in sync?
Freshness depends on the source. Database queries can run on demand; Google Sheets and other snapshot connectors need a successful refresh. Uploaded Excel, CSV, and Tally data stays at the imported version until updated.
Can I ask across all sources at once?
Yes. Ask a single question in natural language and tableArth.ai pulls data across every source connected to the Workbook to return one unified answer.