Document Upload & Management

Complete guide to managing your knowledge base documents

Document Management Overview

The platform supports comprehensive document management with intelligent processing, metadata enrichment, and advanced search capabilities. Documents are processed using AI to extract meaning and create searchable knowledge bases.

Smart Processing

AI-powered document analysis and chunking

Rich Metadata

Comprehensive tagging and categorization

Intelligent Search

Semantic search and AI-powered queries

Supported File Types

File formats that can be uploaded and processed

✅ Fully Supported

PDF Documents

Full text extraction, images, tables

Microsoft Word (.docx)

Text, formatting, embedded content

Plain Text (.txt)

Raw text content

Markdown (.md)RECOMMENDED

Optimal format for AI processing

⚠️ Limited Support

PowerPoint (.pptx)

Text extraction only

Excel (.xlsx)

Text content, limited formatting

RTF Documents

Basic text extraction

Note: File size limit is configurable per workspace (default: 50MB)

🎵 Media Files (Transcription Service)

Video FilesUsage Balance Required

MP4, MOV, AVI, MKV - Transcribed to markdown

Audio FilesUsage Balance Required

MP3, WAV, M4A, FLAC - Transcribed to markdown

Transcription Service: Audio and video files are automatically transcribed into searchable markdown format. Charges apply based on content duration.

Convert Files for Upload

Transform your files into the optimal format for enhanced AI performance

Recommended: Markdown (.md) Format

We strongly recommend converting your documents to Markdown format before uploading. Markdown files provide the highest quality results with our Smart Language Model (SLM) and deliver superior search accuracy and AI responses.

Why Markdown Works Best:

Clean, structured text without formatting artifacts
Preserves document hierarchy and relationships
Optimal chunking and context preservation
Enhanced AI understanding and citation accuracy

Performance Benefits:

Faster processing and indexing
More accurate search results
Better AI response quality
Improved citation precision

Automatic File Conversion

How It Works

Our conversion service automatically transforms your documents into optimized markdown format, preserving structure while removing formatting noise that can interfere with AI processing.

Preserves headings, lists, and tables

Removes unnecessary formatting

Maintains document structure

Optimizes for AI processing

Supported Conversions

PDF → Markdown

Word (.docx) → Markdown

PowerPoint (.pptx) → Markdown

HTML → Markdown

RTF → Markdown

Audio & Video Transcription ServiceRequires Usage Balance

Transform your audio and video content into searchable, AI-ready text. Our transcription service converts spoken content into high-quality markdown documents that integrate seamlessly with your knowledge base.

Supported Media Types

Audio files (MP3, WAV, M4A, FLAC)
Video files (MP4, MOV, AVI, MKV)
Meeting recordings
Training videos and webinars
Interviews and presentations

Usage & Benefits

High-accuracy speech recognition
Automatic speaker identification
Timestamp preservation
Formatted markdown output
Charged per minute of audio/video

Usage Balance Required: Transcription services consume usage balance based on the length of audio/video content. Check your workspace balance before uploading large media files.

Why Use File Conversion & Transcription?

Enhanced Accuracy

Markdown format provides cleaner text that AI can process more accurately, leading to better search results and more precise answers.

Faster Processing

Optimized formats process faster and consume fewer resources, making your knowledge base more responsive and efficient.

Better Context

Structured markdown preserves document hierarchy and relationships, enabling more contextual and relevant AI responses.

Upload Methods

Different ways to add documents to your workspace

Single File Upload

Upload individual files with detailed metadata configuration.

Drag and drop or click to browse files

Add comprehensive metadata during upload

Configure chunking options for large documents

Real-time processing status and feedback

Batch File Processing

Upload multiple files simultaneously with shared metadata templates.

Select multiple files at once

Apply metadata templates to all files

Queue management with progress tracking

Error handling and retry mechanisms

Metadata Management

Organize and categorize your documents effectively

Core Document Information

Document ID

Unique identifier (auto-generated if not provided)

Document Title

Human-readable title for the document

Author

Document creator or responsible person

Version

Document version (default: 1.0)

Description

Brief summary of document content

Organizational Categories

Department

• Engineering

• Marketing

• Sales

• Finance

• HR

• Legal

• Operations

Document Type

• Manual

• Report

• Policy

• Contract

• Guide

• Analysis

• Specification

Priority Level

• High

• Medium

• Low

Business Context

Business Unit

Organizational unit or division

Project

Associated project or initiative

Cost Center

Financial tracking code

Confidentiality

Public, Internal, Confidential, Restricted

Custom Metadata SchemaWorkspace Customizable

Every workspace can define its own custom metadata schema tailored to your organization's specific needs. This powerful feature allows you to create structured data fields that enhance AI filtering and improve response relevance for your unique use cases.

Customization Benefits

Filter responses by your organization's structure
Create industry-specific categorization
Implement compliance and governance tags
Enable role-based content filtering
Support multi-language or regional variants

Enhanced AI Filtering

More relevant search results
Context-aware AI responses
Automated content routing
Precision document discovery
Intelligent content recommendations

Industry-Specific Examples

Healthcare

• Patient Category (Adult, Pediatric, Geriatric)

• Medical Specialty (Cardiology, Oncology, etc.)

• Compliance Level (HIPAA, FDA, Clinical Trial)

• Treatment Phase (Diagnosis, Treatment, Follow-up)

• Evidence Level (Research, Guidelines, Protocol)

Legal

• Practice Area (Corporate, Litigation, IP)

• Jurisdiction (Federal, State, International)

• Document Status (Draft, Final, Archived)

• Client Privilege (Privileged, Work Product)

• Matter Type (Transactional, Regulatory)

Manufacturing

• Product Line (Electronics, Automotive, etc.)

• Quality Standard (ISO, Six Sigma, Lean)

• Safety Classification (OSHA, Environmental)

• Production Stage (Design, Testing, Production)

• Supplier Category (Tier 1, Tier 2, Critical)

Financial Services

• Product Type (Banking, Insurance, Investment)

• Risk Level (Low, Medium, High, Critical)

• Regulatory Framework (SOX, Basel III, MiFID)

• Client Segment (Retail, Corporate, Institutional)

• Geographic Region (Domestic, EMEA, APAC)

Education

• Grade Level (K-12, Undergraduate, Graduate)

• Subject Area (STEM, Liberal Arts, Vocational)

• Learning Objective (Knowledge, Skills, Assessment)

• Accessibility (ADA Compliant, Multi-language)

• Content Type (Curriculum, Assessment, Research)

Technology

• Technology Stack (Frontend, Backend, DevOps)

• Development Phase (Planning, Development, Production)

• Security Level (Public, Internal, Confidential)

• API Version (v1, v2, Beta, Deprecated)

• Platform (Web, Mobile, Desktop, Cloud)

Setting Up Custom Metadata

Configuration Steps

1Access workspace settings as an administrator
2Navigate to "Metadata Schema" configuration
3Define custom field names and types
4Set validation rules and default values
5Apply schema to new and existing documents

Best Practices

Start with essential fields, expand gradually
Use consistent naming conventions
Provide clear field descriptions for users
Test filtering scenarios before full deployment
Train users on metadata importance and usage

Document Chunking

Breaking large documents into searchable segments

What is Document Chunking?

Large documents are automatically split into smaller, manageable segments (chunks) that can be independently searched and referenced by the AI. This improves search accuracy and enables precise citations.

Chunking Options

Chunk Size

Number of characters per chunk (default: 2048)

Recommended: 1024-4096 characters

Overlap Size

Characters overlapping between chunks (default: 200)

Recommended: 10-20% of chunk size

Best Practices

Use smaller chunks (1024-2048) for detailed search

Use larger chunks (3072-4096) for context preservation

Maintain 10-20% overlap to preserve context

Test different sizes based on document type

Document Management Actions

Available operations for managing uploaded documents

Viewing & Navigation

View Document

Browse document content and chunks

Search Within

Find specific content in the document

Download

Download original file

Management Actions

Edit Metadata

Update document information and tags

Reprocess

Re-chunk with different settings

Delete

Remove document and all chunks

Common Issues & Solutions

Upload Fails or Times Out

Check file size limits, internet connection, and file format compatibility. Large files may need to be split or compressed.

Poor Search Results

Add more descriptive metadata, use appropriate chunking settings, and ensure document content is text-searchable (not image-only PDFs).

Processing Takes Too Long

Large documents with complex formatting may take several minutes to process. Check the processing queue and be patient with large files.

Next Steps

After uploading documents, explore these features