Overview
Harvey allows you to upload files to Assistant and Vault to analyse documents, extract structured information and collaborate with others.
This article explains:
- Which file types are supported
- The maximum processable file size for each format
- Additional processing and capacity limits
- How to upload files to Assistant and Vault
- How to troubleshoot common upload issue
Supported file types and size limits
| File type | Maximum processable file size | Additional notes and limits |
|---|---|---|
| Portable Document Format (.pdf) | 500 MB | Text extraction is limited to 10,000 pages. PDFs containing more than 10,000 pages will fail to process when text extraction is required. |
| Word (.doc and .docx) | 500 MB | Comments within Word documents are extracted and added to the footer. |
| PowerPoint (.ppt and .pptx) | 500 MB | No additional limits specified. |
| Excel (.xls and .xlsx) | 20 MB | Maximum of 2 million cells. Files containing more than 2 million cells will fail to process. |
| Email (.eml and .msg) | 500 MB | Email attachments are automatically extracted. |
| Microsoft Outlook Personal Storage Table (.pst) | 500 MB | Emails within the PST file are extracted and added to a new folder in Vault. Extracted files count towards the Vault project’s file limit. |
| Comma-Separated Values (.csv) | 20 MB | Maximum of 2 million cells. Files containing more than 2 million cells will fail to process. |
| Text (.txt) | 30 MB | No additional limits specified. |
| HTML | 8 MB | No additional limits specified. |
| Images (.jpg, .png and .tiff) | 500 MB | Harvey extracts text from images. Charts and diagrams are not fully supported. |
| Common code and markup languages | 15 MB | Includes PHP, Python, Python Script, Java, JavaScript, TypeScript, JSON, XML, C, C++, CSS, Ruby, TeX and Markdown. |
| Audio (.m4a, .mp3, .wav, .webm, .flac and .ogg) | 500 MB in Assistant; 4 GB in Vault | Maximum duration is 2 hours per file. Audio is automatically transcribed into a Word document with speaker labels, timestamps and the automatically detected language. |
| Relativity Short Message Format (.rsmf) | 50 MB | Messages are extracted into individual files. |
| ASiC-E or DigiDoc archive (.asice) | 4 GB | Can only be uploaded to Vault. If uploaded to Assistant, the file will be rejected and the user will be directed to Vault. |
| HTTP Archive (.har) | 500 MB | Can only be uploaded to Vault. The file is extracted and divided into separate page-level documents. |
Assistant and Vault capacity limits
- Assistant allows up to 50 files per thread.
- A Vault can store up to 100,000 files.
- Files extracted from formats such as PST and RSMF count towards the Vault project’s file limit.
- Before uploading an archive containing many items, check that the Vault has sufficient available file capacity.
Upload files to Assistant
- Open a new Assistant thread.
- Enter your prompt.
- Select Files.
- Upload files from your computer, document management system or an existing Vault.
- If required and available, select additional knowledge sources under Sources.
- Select Ask Harvey.
You can add or replace files during the thread and continue asking follow-up questions
Tips for successful file processing
- Use clean PDF or DOCX files for better text extraction.
- Ensure PDFs requiring text extraction do not exceed 10,000 pages.
- Ensure Excel and CSV files do not exceed 20 MB or 2 million cells.
- Check the available Vault file capacity before uploading a PST file.
- Remember that a PST file can contain thousands of emails, with each extracted email counting towards the Vault file limit.
- Use Vault for large document sets that need to be organised and reused.
- Be aware that charts and diagrams in image files may not be fully processed.
- Check the 15 MB limit before uploading code or markup files.
- Ensure audio files do not exceed 2 hours.
- Confirm that Vault-only formats are not being uploaded to Assistant.
Frequently asked questions
What is the maximum file size in Harvey?
The maximum size depends on the file type and upload location. Many common document formats support files up to 500 MB, while some formats have lower limits.
Audio files can be up to 500 MB in Assistant or 4 GB in Vault. ASiC-E files can be up to 4 GB and can only be uploaded to Vault.
Do the limits apply to each individual file?
Yes. The limits apply to each individual file uploaded to Assistant or Vault.
Do these limits apply to Workflows?
No. Workflows use their own upload process and limits.
How many files can I upload to an Assistant thread?
Assistant supports up to 50 files per thread.
How many files can a Vault store?
A Vault can store up to 100,000 files. Files extracted from archives and message formats may count towards this limit.
Can I upload an Excel or CSV file larger than 20 MB?
No. Excel and CSV files must not exceed 20 MB or 2 million cells. Both limits apply.
Can I upload a PDF containing more than 10,000 pages?
A PDF requiring text extraction will fail to process if it contains more than 10,000 pages.
Can I upload audio files?
Yes. Supported formats include M4A, MP3, WAV, WebM, FLAC and OGG.
Audio files can be up to 500 MB in Assistant or 4 GB in Vault. The maximum duration is 2 hours per file.
What happens after an audio file is uploaded?
Harvey automatically transcribes the audio into a Word document with speaker labels, timestamps and the automatically detected language.
Are charts and diagrams extracted from images?
Harvey can extract text from supported image files, but charts and diagrams are not fully supported.
Should I use Assistant or Vault?
Use Assistant for individual requests or a small set of files.
Use Vault when working with larger document sets, matter-level work or files that need to be organised and reused.
Article maintenance notice
Harvey is evolving quickly, and its supported file types, processing limits and interface may change. This Knowledge Base article should be reviewed frequently against Harvey’s official documentation to ensure the information remains accurate and current.