Research inputs: supported files and sources
Praxiom AI accepts research as uploaded files or as imports from Google Drive. You can upload transcripts, audio and video recordings, PDFs, Word and PowerPoint files, spreadsheets, CSV and JSON exports, and images, up to 100 MB per file. From Drive you can import single files or whole folders, including subfolders and .zip archives.
- Updated
Which file types can you upload?
| Kind | Formats | How Praxiom AI reads it |
|---|---|---|
| Documents | PDF, Word (.docx, .doc), PowerPoint (.pptx), EPUB, RTF | Extracts the text; images embedded in PDF, Word and PowerPoint files are described too |
| Text | Plain text, Markdown, HTML | Reads the text, detecting the encoding |
| Spreadsheets and data | CSV, Excel (.xlsx, .xls), JSON | Reads rows and fields as structured text, with column names |
| Transcripts | WebVTT (.vtt), SubRip (.srt) | Parses the timed captions directly, without an AI model |
| Audio | MP3, WAV, M4A, OGG, FLAC | Transcribes speech to text |
| Video | MP4, WebM | Transcribes the audio track |
| Images | PNG, JPEG, GIF, WebP | Describes the image with a vision model |
Executables and scripts are always rejected, whatever their extension. If text extraction fails or finds nothing usable, the source is marked failed with a reason instead of being analyzed as empty.
What are the size limits?
| Limit | Value |
|---|---|
| Direct upload | 100 MB per file |
| Audio and video transcription | 25 MB per file; split longer recordings |
| CSV files | First 10,000 rows |
| Google Drive download (non-Google files) | 25 MB per file |
| Text kept per Google Drive file | 100,000 characters |
| Google Drive folder import | Up to 200 files and 8 subfolder levels per import |
| .zip archive contents | Up to 400 files listed; up to 20,000 characters per file and 200,000 in total |
| New research sources on Pro | 50 per month |
| New research sources on Growth | 200 per month |
| New research sources on Power | 500 per month |
How do you import research from Google Drive?
- Connect Drive from Settings, then Integrations (see Google Drive integration).
- In the Research view, click Import from Drive.
- Search or browse, then select a file or a folder. Shared drives are included.
- Choose the source type and import. A folder is expanded into one research source per file.
Google Docs keep their headings, Google Sheets are read as tables, and Google Slides are read as text. In a folder import, images, audio and video are skipped so a large folder does not trigger hundreds of transcriptions; import those files one at a time instead. .zip archives are expanded and the text files inside them (notes, exports, code) are read, while dependency and build folders are skipped.
What source types does Praxiom AI use?
Every source has one of six types. You choose it when you import from Drive; for uploads Praxiom AI assigns one from the file, and the type stays attached to the source.
- Interview: customer or user interview transcripts and recordings.
- Support ticket: tickets and help desk exports.
- Survey: survey responses and NPS verbatims.
- Usage data: product analytics exports.
- Feedback: feature requests, reviews and other feedback.
- Other: anything else, such as strategy docs or notes.
What happens after you add a source?
- The file is stored privately in your workspace and a research source is created with status pending.
- Praxiom AI extracts the text in the background: parsing documents, transcribing recordings, describing images.
- The source moves to completed, or to failed with the reason shown.
- Completed sources can be selected for a synthesis. See Research Quality Score for how results are scored.
Frequently asked questions
Can Praxiom AI transcribe interview recordings?
Yes. Upload MP3, WAV, M4A, OGG, FLAC, MP4 or WebM files up to 25 MB and Praxiom AI transcribes them. For longer recordings, split the file or upload the transcript from your recording tool (.vtt, .srt, .txt or .docx).
Can I upload a CSV of support tickets?
Yes. Praxiom AI reads each row with its column names, up to 10,000 rows per file, so help desk exports work without reformatting.
Can I import a .zip file?
Yes, through Google Drive. Praxiom AI expands the archive and reads the text files inside it, skipping binaries and dependency folders.
Does uploading research use credits?
No. Uploading and extracting text does not use credits; it counts toward your plan's monthly research source allowance. Running a synthesis on the sources uses credits.
What happens if a file cannot be read?
The source is marked failed with a reason, such as no extractable text, and it is left out of syntheses. Nothing is analyzed as if it were empty.
Run your product discovery on autopilot.
Bring your interviews, tickets and surveys. Get scored insights, a PRD and tickets.