Structured Data
Import catalogs, menus, and price lists as rows, and use live lookup only for connected current data
Structured Data#
Structured data is a table in Agent Studio → Data sources. Each row is one product, service, menu item, or other record. It is not a document, a Q&A, or a dated announcement.
An uploaded spreadsheet is a snapshot. A connected Google Sheet or API can be refreshed on a schedule. Current stock, orders, or availability still need Live lookup on that connected source. A lookup reads matching rows. It does not reserve inventory or create a booking.
Where the rows come from#
| Source | What it is |
|---|---|
| File upload | A static spreadsheet snapshot. It cannot turn on live lookup. |
| Google Drive | A connected spreadsheet that can be refreshed. |
| API | A connected endpoint that returns rows and can be refreshed. |
Use one table per sheet, with a single header row and no merged cells. Confirm the column names and types before relying on the import. The setup screen shows how large a catalog can be imported.
Add a table#
- Open Agent Studio → Data sources.
- Choose Add source, then Structured data.
- Choose File Upload, Google Drive, or API.
- Map each column to the field it represents, such as name, price, description, or category.
- For a Google Sheet or API, set the sync schedule and turn on Live lookup if answers must use current rows.
- Save and wait until preparation finishes.
- Test a question that names a real item in the table.
Structured data does not use the same Train agent step as a file or a website. Follow the status below.
Status#
These are the Data sources statuses. The import screen can show the finished table as Trained. That is the same finished state as Ready in Data sources, not a second lifecycle.
| Status | What to do |
|---|---|
| Setup incomplete | Finish the connection or column mapping. |
| Syncing | Rows are being imported. Wait. |
| Training | Imported rows are being prepared for answers. |
| Ready | The table can be used. An upload is still a snapshot of the file you imported. |
| Review changes | The sheet or API shape changed. Review the columns before trusting new answers. |
| Needs attention | The sync or preparation failed. Use the offered retry, or reconnect the source. |
Refresh schedule#
Connected sheets and APIs can stay on Manual, or refresh Hourly, Every 6 hours, Daily, or Weekly. The schedule updates the imported rows. Live lookup can still read returned values when a customer asks. Uploaded files stay manual. Re-upload the spreadsheet when that snapshot should change.
Questions a table can answer#
| Column | Example |
|---|---|
| Text | "Show the Tokyo packages." |
| Number | "Which packages are under ₱40,000?" |
| Date | "Which trips include December 5?" |
| Yes/No | "Which packages are still available?" |
Keep one record per row, keep numbers numeric, and use a consistent date format. Correct a detected column type in the source configuration if it is wrong.
In a Procedure, add @Look up live data and set its conditions from details the Procedure collected. In a Workflow, add Run an action, choose Live Data Lookup, and set whether every condition must match or any one condition is enough. Later steps can use the matches that action saved.