Exporting Bulk Job Results
Once a bulk job is complete — or even partially complete — you can download all results as a CSV or XLSX file.
How to download
- Open the bulk job from the Bulk Processing list.
- Click the Download button in the top right corner.
- Select CSV or XLSX from the dropdown.
- The file downloads immediately to your browser's downloads folder.
The download button is enabled as soon as at least one row has completed — you do not need to wait for the entire job to finish.
What's included in the export
The export contains every row from your original CSV, plus the AI-generated output columns:
| Column type | Description |
|---|---|
| Original columns | All columns from your uploaded CSV, unchanged |
| Output column(s) | The AI-generated result for each row |
| Status | completed, failed, or pending for each row |
| Error | For failed rows, the error message explaining what went wrong |
| Processing time (ms) | How long the row took to process |
Failed rows are included in the export with their original data and an error message in the error column. This makes it easy to identify and fix the problem rows, then re-process them in a new job.
CSV format
The CSV export is UTF-8 encoded with comma delimiters and standard quoting for values that contain commas or line breaks. It opens cleanly in Excel, Google Sheets, or any standard spreadsheet application.
XLSX format
The XLSX export is a formatted Excel spreadsheet with:
- A header row with bold styling
- Colour-coded status column: green for completed, red for failed, grey for pending
- Column widths auto-sized to fit content (up to a maximum)
XLSX is the better choice if you plan to share the results with stakeholders who will view it directly in Excel or Google Sheets.
Re-processing failed rows
There is no built-in "retry failed rows" button. To re-process failed rows:
- Download the CSV export.
- Filter the CSV to show only rows where
status=failed. - Remove the
status,error, andprocessing_time_mscolumns. - Upload this filtered file as a new bulk job with the same configuration.
Re-processing consumes credits for the failed rows again. Review the error messages first to understand why rows failed — if the cause is a configuration issue (wrong column mapped, model error), fix the configuration before re-running.
Was this helpful?