Upload a CSV
Add a CSV file to AskPhi by dragging it into the upload card or browsing for it.
Before you start
- A
.csvfile with a header row. - Up to 10,000 data rows (header not counted).
- The file should be encoded in UTF-8 (most spreadsheet exports are).
If you are not sure your file qualifies, see System requirements.
Steps
- Open the Chat home (click the AskPhi logo, or + New Chat in the sidebar).
- Find the Upload New Data Source card on the right side.
- Either:
- Drag and drop the CSV onto the dashed drop zone, or
- Click the drop zone and pick a file from your computer.
- The file appears in a preview row showing its name and size.
- Click Upload File.
- Watch the Uploading: x% progress under the file name. When it reaches 100%, a green "File uploaded successfully!" toast appears.
The newly uploaded file becomes the selected file. The right side of the chat home turns into a Create Session form so you can chat with it right away.
Why uploads happen in chunks
AskPhi splits your file into 5 MB pieces and sends them one after another. This:
- Avoids browser timeouts on slower connections.
- Lets the upload bar update smoothly as each chunk completes.
- Reduces the chance of having to retry a whole large file.
You don't have to do anything — chunking is automatic.
Replacing or re-uploading a file
AskPhi treats every upload as a new file. To replace data, you can either:
- Upload again with a new name, then delete the old file from Data Sources, or
- Open the existing file's data editor and use Append CSV to merge new rows in.
Troubleshooting
- "Only .csv files are allowed." — pick a file with a
.csvextension. Excel.xlsxis not supported. Save your sheet as CSV first. - Upload stops mid-way — your network may have dropped. Click Upload File again; AskPhi will start fresh.
- Upload completes but my file isn't in the list — refresh the page. The Data Sources list updates after a successful upload, but a hard refresh always reflects the latest state.
- "Upload failed. Please try again." — usually the file is too large (> 10,000 data rows) or the server is busy. Try a smaller subset or wait a minute.