Why Use a Knowledge Base?
A knowledge base lets you work with multiple files simultaneously, retrieving information and gaining insights across all your documents at once. Your AI agent acts as a mediator, providing relevant answers with sources from your centralized data.
Common use cases:
- Connect Google Drive files with one click
- Summarize research papers or meeting notes
- Extract key information from PDFs and Word documents
- Analyze data in CSV sheets
- Access publicly available web URLs
Department-Specific AI Agents
Knowledge bases enable powerful enterprise scenarios where different AI agents access specific company knowledge for targeted purposes:
Sales & Customer Success:
- AI agent connected to product documentation, pricing sheets, and case studies
- Instantly generates personalized proposals and answers client questions
- Accesses competitor analysis and market research for strategic conversations
HR & People Operations:
- AI agent with access to employee handbooks, policies, and compliance documents
- Automates onboarding processes and policy clarifications
- Provides instant answers to benefits, leave policies, and company procedures
Legal & Compliance:
- AI agent trained on contracts, regulatory documents, and legal precedents
- Reviews contracts for compliance issues and standard clause variations
- Provides quick access to regulatory requirements and internal legal guidance
Product & Engineering:
- AI agent connected to technical documentation, API references, and bug reports
- Assists developers with code examples and troubleshooting
- Maintains knowledge of system architecture and deployment procedures
Marketing & Content:
- AI agent with brand guidelines, campaign data, and market research
- Creates consistent messaging across channels using approved content
- Accesses performance metrics and campaign histories for optimization
How Knowledge Bases Work
When you query your knowledge base, your AI agent searches for relevant sections matching your intent, similar to a search engine. Instead of processing everything at once, it retrieves only the sections related to your question, making results faster and more accurate.
Key limitation: AI models have a limited "context window" (character processing capacity). Being specific in your queries improves results significantly.
Search Tips & Best Practices
- Broad searches: Find relevant documents across a wider topic / several knowledge bases
- Focused searches: Get detailed answers from specific files once you've identified them
- Keywords: Search for specific terms, definitions, or documentation
Multi-Document Queries
Yes, you can compare and extract connections between different documents. However, avoid querying too many files with large amounts of text at once - split complex tasks into smaller sub-tasks instead.
Structuring Your Knowledge Base
Our recommendation: Group knowledge bases by purpose or department (e.g., "HR Policies," "Marketing Assets") rather than creating one massive repository.
Benefits of focused KBs:
- Easier sharing and access control
- Better AI accuracy and precision
- More efficient use of context window
- Department-specific AI agents can access only relevant information
- Reduced security risks through targeted data access
Advanced Enterprise Setup
Multi-Agent Architecture:
- Deploy specialized AI agents for each department
- Connect agents to relevant knowledge bases through integrations
- Maintain data security while enabling cross-functional collaboration
- Scale AI assistance across the entire organization
Integration Examples:
- Sales team using chat interface with AI agents connected to sales materials
- Support team accessing AI agents trained on troubleshooting guides (i.e. FAQs, help center articles) and product documentation
- HR team chatting with AI agents that have access to employee handbooks and policy documents
- Engineering team using AI agents connected to technical documentation
Supported File Types & Limitations
Supported formats:
- Text: .txt, .md, .html
- Documents: PDF, Word (.docx), PowerPoint (.pptx)
- Image: .png, .jpg (can only be uploaded in chat)
- Data: Excel (.xml, .xlsx), CSV
- Integration-specific sources: Web URLs, SharePoint Sites/Pages (.aspx), Google Docs, Google Sheets..
Size limits:
- Regular files: 300MB maximum per file
- Tabular data (Excel, CSV): 10 million character limit across all rows, columns, and sheets combined
Content density considerations:
Even files under 300MB may fail to upload if they contain extremely dense text content that exceeds processing limits. Files with dense formatting, extensive metadata, or highly compressed text (such as PDFs with small fonts or tables with thousands of entries) can sometimes hit character processing limits before reaching the file size limit.
If you encounter upload issues with smaller files, try:
- Breaking large documents into smaller sections
- Simplifying formatting or removing unnecessary metadata
- For spreadsheets, splitting data across multiple files if row/column count is very high
Getting Started
- Go to the "Knowledge" section in the TextCortex sidebar
- Click "+New" and name your Knowledge Base
- Upload your files
- In Chat, click "Add Context" and select your Knowledge Base
- Start searching and working with your documents
Note: All files are processed by TextCortex without third-party involvement. See our Trust Center for more details.