Gen 3: This is a legacy Gen 2 article. For current GT AI OS 3.0 guidance, see gen3/datasets/managing.

Managing Documents

This guide covers how to add, manage, and organize documents within datasets.

Adding Documents

Uploading Files

1. Open the dataset details page
2. Click Upload Documents or drag files to the upload area
3. Select files from your computer
4. Wait for processing to complete

Supported File Types

Type	Extensions	Notes
PDF	.pdf	Best for formatted documents
Word	.docx	Standard office documents
Text	.txt	Plain text files
Markdown	.md	Documentation, notes
CSV	.csv	Tabular data
JSON	.json	Structured data

Upload Limits

• Maximum file size: 50MB per file
• Large files may take longer to process

Document Processing

What Happens During Processing

When you upload a document:

1. Extraction: Text is extracted from the file
2. Chunking: Text is split into searchable segments
3. Embedding: Each chunk is converted to a vector
4. Indexing: Vectors are stored for fast retrieval

Processing Status

Documents show their status:

• Pending: Waiting to upload
• Uploading: File transfer in progress
• Processing: Being analyzed and embedded
• Completed: Available for use
• Failed: Error during processing

Managing Documents

Viewing Document Details

Click on a document to see:

• File name and type
• Upload date
• Processing status
• Chunk and vector counts

Removing Documents

To delete a document:

1. Find it in the document list
2. Click the delete icon
3. Confirm removal

Note: Removal triggers reindexing of the dataset.

Managing Multiple Documents

Documents are managed individually within a dataset. For bulk operations at the dataset level, use the Export and Import features on the Datasets page.

Document Quality

Best Practices for Source Documents

Content quality:

• Use clear, well-structured text
• Avoid scanned images (OCR limitations)
• Include context and definitions

File quality:

• Use native text formats when possible
• Ensure files aren't corrupted
• Keep file sizes reasonable

Optimizing for Search

Help agents find relevant content:

• Use descriptive headings
• Include key terms naturally
• Organize content logically

Troubleshooting

Document Not Processing

If processing gets stuck:

1. Check file format is supported
2. Verify file isn't corrupted
3. Try uploading a smaller version
4. Contact support if issue persists

Content Not Found in Searches

If expected content isn't retrieved:

1. Verify document processed successfully
2. Check content isn't in images (not searchable)
3. Try searching with exact phrases
4. Consider document chunking settings

Poor Quality Results

If search results aren't relevant:

1. Review document content quality
2. Consider splitting large documents
3. Update chunking configuration
4. Add more context to documents