Ask about data and pull it into Excel

DataMind Oda answers questions about your company's data through the Minds of your DataMind OS. This page follows the journey from the question you type in the pane to the new sheet the result lands in. See Oda in Excel for the pane itself.

Ask a question about your data

1. Type your question in the message box, in plain language, for example Show this year's sales by region. The hint Ask about your data or this workbook… stays there until you type. Enter sends the message, or click Send; Shift+Enter starts a new line.
2. Oda asks the Mind whose name and description fit the question, or the Mind you set in the Asking: chip; see Choosing the Mind. A step row names the work as it happens, such as Asking the Mind.
3. The answer appears in the chat; it can take minutes, and it names the queries the Mind ran.

Follow-up questions in the same chat continue the conversation with that Mind, also after you close and reopen the pane.

If the Mind needs more detail, Oda asks you its question in the chat and waits. Answer in your next message; Oda passes your answer back to the Mind and shows the finished answer.

Asking Minds needs the Use DataMind Oda and Use Minds from DataMind Oda permissions. Without Use Minds from DataMind Oda, the pane hides the Asking: chip and works with the workbook only.

Pull a Mind's result into a new sheet

When you want the rows themselves in the workbook, ask for the full result in a sheet, for example Pull those numbers into a sheet. Oda runs the query behind the answer again on your DataMind server and writes its full result, so the sheet holds the data as it is at the moment of the pull. If the data changed after the Mind answered, the sheet can differ from the numbers in the answer.

1. Oda adds a new sheet to the workbook, writes the result as a formatted Excel table with a header row, fits the column widths and switches to the new sheet. The step row reads Pulling data into a new sheet, then the number of rows and the sheet.
2. The sheet is named after the query, or after the name you ask for in your message. A long name is shortened to at most 28 characters, and if the workbook already has a sheet with that name, Oda adds a number, such as Sales 2.

• Adding the sheet follows your permission mode: in Ask before edits, a card asks first; in the other modes the pull runs on its own. See Permission modes and approvals in Excel.
• One pull writes at most 100,000 rows; a bigger result is cut there. Narrow the question first and pull that.
• If the query takes longer than 15 minutes to export, the pull stops before anything is written, and the step row reads This took too long and was stopped.
• Oda pulls only queries from Mind answers in the same chat. In another chat, ask the Mind again.
• If you click Stop while the rows are being written, Oda removes the new sheet again, so the workbook stays as it was.

For everything else Oda can see and change in the workbook, see How Oda works in your workbook.

Text, numbers and dates in the new sheet

Oda writes each column by the type the Mind's query reports for it, so nothing changes under you:

• Text columns stay exactly as stored: codes with leading zeros such as 00123, long identifiers, codes that look like dates or numbers, and text that starts with =, +, - or @, which never becomes a formula.
• Number columns land as numbers you can calculate with. A number with more than 15 digits before the decimal point, such as a 20-digit account number, stays text, because Excel keeps only 15 digits of a number.
• Date and date-time columns land as real Excel dates, shown as yyyy-mm-dd or yyyy-mm-dd hh:mm:ss, so you can sort, filter and calculate with them. Dates before 1 March 1900 stay text, and so do all dates on a version of Excel that cannot tell which date system the workbook uses.
• Long text lands in full up to 32,767 characters, the most one Excel cell holds; anything beyond that is cut.
• Empty values stay empty.

When the query reports no type for a column, plain figures become numbers and everything else, including codes with leading zeros, stays text.

Where the data goes

• Oda sends the Mind your question, not your workbook.
• The pulled rows go straight from the query's result into the sheet. The chat keeps only the sheet name, the table name, the column names and the row count, never the rows.
• The workbook keeps a short note of each pull: the sheet and table that received the data, the Mind and the query behind it, the row count and the time. The notes hold names and numbers only, never SQL and never the data. The most recent 200 notes are kept for a workbook; beyond that, the oldest drop off.

When a Mind job does not finish

Sometimes a Mind job stops before it answers. The step row reads This step did not work., and Details shows the reason:

• The Mind could not finish: <reason>. The Mind stopped with an error; <reason> is what the Mind reported, or no reason given when it reports nothing.
• The Mind job was cancelled. The job stopped before it finished.
• The Mind did not finish within an hour, so the job was cancelled. A job that runs for more than an hour is cancelled.
• DataMind answered HTTP <status>. The request to your DataMind server failed; <status> is the code it returned. When the server sends its own explanation, that is shown instead.

A job can also be refused before it starts: Oda tells you when too many DataMind jobs are open for your account, or when jobs are being started too quickly. Wait for your running jobs to finish, then try again.

Ask the question again. If the same message keeps returning, contact your administrator with it.

Questions, answered

How long does a Mind's answer take?

Answers can take minutes. A job that runs for more than an hour is cancelled, and the chat says so.

What happens if the Mind asks me something back?

Oda asks you the Mind's question in the chat and waits. Answer in your next message; Oda passes your answer back to the Mind and shows the finished answer.

Why do the numbers in the sheet differ from the Mind's answer?

A pull runs the query again and writes the data as it is at that moment. If the data changed after the Mind answered, for example because a load ran in between, the sheet shows the newer numbers.

Can I pull several results from one answer?

Yes. The answer names the queries the Mind ran; ask for the one you want in a sheet. Each pull adds its own new sheet.

Why did a number land in my sheet as text?

The query reports that column as text, or the value is not a plain figure: codes with leading zeros and numbers with more than 15 digits stay text, so nothing gets damaged. Number columns land as numbers you can calculate with.

Do dates land as real dates?

Yes. Date and date-time columns land as Excel dates you can sort and calculate with. Dates before 1 March 1900 stay text, and so do all dates on a version of Excel that cannot tell the workbook's date system.

Can I pull more than 100,000 rows?

No. A single pull writes at most 100,000 rows, and a bigger result is cut there. Narrow the question first and pull that.

Can I name the new sheet myself?

Yes. Say the name in your message and Oda uses it. A long name is shortened to at most 28 characters, and if a sheet already uses the name, Oda adds a number, such as Sales 2.

Does the chat keep the rows Oda pulled?

No. The rows go into your sheet; the chat keeps only the sheet name, the table name, the column names and the row count. The workbook's note about the pull holds names and numbers only.

Why could Oda not pull a Mind's result?

Minds that run queries as each user cannot have their results pulled into a sheet. Oda says Exporting is not available yet for Minds that run queries as each user. Use a preview query instead. Ask the Mind for a preview of the result instead. A pull also needs a query from a Mind answer in the same chat.

Does a pull change anything I already have?

No. A pull adds a new sheet; your existing sheets keep their contents.