Knowledge Base

Train your AI agent with your business documentation.

What is the Knowledge Base?

The knowledge base is your AI's reference library:

Customer Question
       ↓
AI searches knowledge base
       ↓
Finds relevant information
       ↓
Generates accurate response

This is called RAG (Retrieval Augmented Generation).

Adding Content

Supported Formats

Format	Extensions
Documents	PDF, DOC, DOCX, TXT
Spreadsheets	XLSX, CSV
Web Pages	URL import
Markdown	MD files
Rich Text	RTF

Upload Documents

1. Go to AI Agents → Knowledge Base
2. Click Upload
3. Select files (up to 50MB each)
4. Wait for processing
5. Documents are indexed

Import from URL

1. Click Import URL
2. Enter the web page URL
3. Fyncall extracts content
4. Content is indexed

Manual Entry

Create knowledge articles directly:

1. Click Create Article
2. Enter title and content
3. Add tags for organization
4. Publish

Organizing Knowledge

Categories

Organize content by topic:

Knowledge Base
├── Products
│   ├── Product A Documentation
│   └── Product B Features
├── Policies
│   ├── Return Policy
│   └── Shipping Policy
└── FAQs
    ├── Billing FAQs
    └── Technical FAQs

Tags

Add tags for fine-grained organization:

[returns] [policies] [30-day]
[shipping] [international] [rates]

Priority

Set content priority:

Priority	Use Case
High	Critical info, always relevant
Normal	Standard content
Low	Supplementary info

Content Best Practices

Write for AI

Structure content clearly:

# Return Policy

## Eligibility
Items can be returned within 30 days of delivery.

## Conditions
- Item must be unused
- Original packaging required
- Receipt or order confirmation needed

## Process
1. Contact support
2. Receive return label
3. Ship item back
4. Refund processed in 5-7 days

Be Specific

Instead of:

"We have a generous return policy."

Write:

"Items can be returned within 30 days for a full refund."

Keep Updated

• Review content regularly
• Update outdated information
• Remove obsolete documents

How AI Uses Knowledge

Semantic Search

AI understands meaning, not just keywords:

Query: "Can I send this back?"
Matches: Return Policy (even without "send back" in text)

Context Assembly

AI combines multiple sources:

Query: "Return my international order"

Sources used:
1. Return Policy (eligibility)
2. International Shipping (process)
3. Refund Policy (timeline)

Response: Combines all relevant info

Citation

AI can cite sources:

"According to our return policy, you have 30 days
to return items in their original condition."

Knowledge Quality

Metrics

Metric	Description
Coverage	% of questions answerable
Accuracy	Correct answer rate
Citation Rate	How often sources used

Improving Quality

Monitor and improve:

1. Review unanswered questions
2. Add missing content
3. Clarify confusing sections
4. Remove duplicate info

Content Sources

Automatic Sync

Sync from external sources:

Source	Sync Method
Help Center	API integration
Notion	Native connection
Google Docs	Import links
Confluence	API key

Keep in Sync

Configure automatic updates:

Sync frequency: Daily
Last synced: 2 hours ago
Next sync: In 22 hours

Testing Knowledge

Ask Questions

Test knowledge coverage:

1. Go to Test Knowledge
2. Ask a question
3. See matched content
4. Verify relevance

Gap Analysis

Find missing topics:

1. Review "No Match" queries
2. Identify patterns
3. Add missing content
4. Re-test