Every system exports one, so every team has dozens: survey exports, transaction extracts, product lists, attendance records. A CSV in a folder is a file you can sort and filter, but not one you can ask a question of — and the moment a column holds a sentence rather than a number, a spreadsheet stops helping. CSV upload is part of askFinz Knowledge, which is still rolling out; here is what it will do, and what it will not.
Ask a dataset a question in words
Upload the file and ask what you actually want to know — "which regions complained about delivery times" — rather than translating it into a filter, a pivot and a formula first.
Search the sentences inside the columns
Free-text columns are where the useful detail hides: comments, reasons, notes. Those are matched by meaning, so a row can come back even when it does not repeat your wording.
Use it beside your other work
An uploaded dataset sits in the same search as your documents and notes, so an answer in Research can draw on the file and cite the row it came from. That matters most when the spreadsheet and the document disagree: both are quoted, and you can see which is which.
What is read, and what is never written
What crosses, and what stays with you
Read
only the file you upload, when you upload it.
Written back
nonenothing. Your original file is untouched.
Stored
a searchable copy of the rows, plus the filename and the column names.
Never read
the rest of your machine, and anything you have not uploaded.
Setting it up
Planned — not available yetWhat connecting it will look like once the connector arrives.
- 01
Drag the file into Knowledge.
- 02
Confirm the columns it detected, and mark any to exclude.
- 03
Choose who in your workspace can see it — just you, or the team.
Limits, honestly
- A CSV is a snapshot. Re-export and re-upload when the source changes; there is no live link.
- Very large files may be split, and extremely wide files may have low-signal columns skipped.
- Formulas are not evaluated — what crosses is the value in the cell.
- Multi-sheet spreadsheets need one export per sheet.
- Unusual delimiters and missing header rows are detected where possible, but a file without headers gives worse answers than one with them.