Knowledge Sources
AI agents rely on structured knowledge sources to provide accurate and consistent answers.
In Flametree, knowledge can be provided through two complementary mechanisms:
- Fast Access Knowledge Base (FAQ) — optimized for short, frequently asked questions
- Vector Knowledge Base — designed for large documents and detailed information
These sources serve different purposes and are typically used together. For the detailed knowledge base tutorials, please refer to the Knowledge Bases section of the documentation.
Knowledge Source Comparison
| Feature | Fast Access FAQ | Knowledge Base |
|---|---|---|
| Typical Content | Common questions | Documentation, catalogs, policies |
| Size Limit | ~10,000 characters | No strict limit |
| Search Method | Embedded in prompt | Vector search (RAG) |
| Tool Required | No | Yes (SingleSearchFAQ) |
| Retrieval Priority | Checked first | Used when additional context is required |
| Best Use Case | Instant, controlled answers | Complex or large datasets |
Fast Access Knowledge Base (FAQ)
The Fast Access FAQ contains predefined question–answer pairs that the agent checks first when responding to customers.
This helps:
- respond faster
- avoid unnecessary or overly long answers
- keep responses consistent and predictable
Example FAQ
## Is your company called Example Industries or Example?
Example Industries is the legal name, while **Example** is the brand name.
## What brands are available in your catalog?
We work with leading brands and update our selection regularly.
## Do you offer delivery?
Yes, you can return or exchange a product within 14 days if it remains in its original condition and packaging.
| Do’s ✅ | Don’ts ❌ |
|---|---|
| - Use short, natural customer questions - Write answers in a conversational tone - Keep FAQ content clear and easy to understand | - Copy formal or legal text directly into FAQ answers - Store frequently changing info in FAQ instead of the Knowledge Base |
Vector Knowledge Base
A Knowledge Base allows the agent to search larger datasets such as manuals, price lists, or structured documentation. The agent retrieves relevant fragments dynamically when answering customer questions.
Knowledge Base search is typically triggered through tools configured in the workflow (for example, SingleSearchFAQ).
Supported File Formats
| Format | Use Case |
|---|---|
| Manuals, reports | |
| DOCX | Documentation |
| TXT | Plain text |
| MD | Structured documentation |
| XLSX | Data tables |
| HTML | Web content |
| JSONL | Structured datasets |
Related Resources
- Knowledge Base — A detailed guide on knowledge bases and how to set up them
- Product Catalogs — An advanced feature to enable AI agent provide detailed product descriptions and make context-aware sales