Data Source — Knowledge Base
A Knowledge Base data source connects your digital employee to an existing question-and-answer knowledge base. Semantic search identifies the questions most relevant to the user's request and passes their content to the AI model to build a response.
This source type is ideal for making the most of an existing FAQ, without having to rewrite its content for the AI.
Configuration
Maximum number of questions to include
Limits the number of questions from the knowledge base passed to the AI model. Too high a value may dilute the relevance of the response; too low may miss useful complementary information.
A value between 3 and 5 is usually a good balance. Increase it if your questions are short and thematically close to each other.
Indexing options
These checkboxes control which elements of each question are included in the semantic search index.
| Option | Description |
|---|---|
| Index the answer | Includes the answer text in the index. Recommended in most cases. |
| Index learn more | Includes "learn more" links and content associated with the question. |
| Index tags | Includes question tags to improve thematic discrimination. |
| Index categories | Includes question categories in the index. |
| Index subcategories | Includes subcategories in the index. |
Enable "Index tags" and "Index categories" if your knowledge base is well structured: this improves the precision of semantic search by adding thematic context.
After changing any indexing option, use "Save and re-index" — not just "Save" — for the changes to take effect in search results.
Filter responses by tag
To restrict this source to a specific context (entity, site, department), fill in the "Filter data sources" section at the bottom of the configuration page. See the Overview for details.