> ## Documentation Index
> Fetch the complete documentation index at: https://docs.nulix.io/llms.txt
> Use this file to discover all available pages before exploring further.

# AI-Powered Semantic Search for Your Knowledge Base

> Enable Nulix AI search to answer natural language questions with direct excerpts — even when readers don't know the exact keywords to search for.

AI search transforms how readers interact with your knowledge base. Instead of requiring readers to guess the right keywords, Nulix AI search understands the intent behind a question and returns a direct excerpt from the most relevant article — making your content genuinely conversational and easier to navigate.

## How AI Search Differs from Keyword Search

Standard keyword search looks for exact or near-exact word matches between a query and your article content. AI search goes further by understanding the **meaning** of a question, not just its words.

| Feature | Keyword Search | AI Search |
| - | - | - |
| Matching method | Exact / fuzzy text match | Semantic (meaning-based) match |
| Query style | Keywords ("password reset") | Natural language ("how do I reset my password?") |
| Result format | List of matching articles | Direct answer excerpt + source article |
| Handles synonyms | Limited | Yes |
| Understands context | No | Yes |

This means a reader can type a full question in plain language and get a pinpointed answer — even if the exact phrase never appears in your article.

## Enabling AI Search

AI search is available on the **Pro plan and above**. Once your workspace is on an eligible plan, enable it for any knowledge base in a few steps.

**1. Open Settings**

In the Nulix sidebar, click **Settings** and select the knowledge base where you want to enable AI search.

**2. Navigate to the Search tab**

Click the **Search** tab in the knowledge base settings panel.

**3. Enable AI Search**

Toggle **Enable AI Search** to the on position and click **Save Changes**. Nulix will begin indexing your articles for semantic search immediately.

> [!NOTE]
>
> Initial indexing for AI search may take a few minutes for large knowledge bases. Readers will see standard keyword results in the meantime, with no downtime or disruption.

## How It Works

When you enable AI search, Nulix processes your published articles and builds a **semantic index** — a representation of the meaning of each section of your content. When a reader submits a query, Nulix compares the meaning of their question against this index to find the best-matching passage, then surfaces that passage as a direct answer above the standard results list.

### Example

A reader searches:

> "how do I reset my password"

Even if your article is titled "Account Security" and uses the phrase "change your credentials" instead of "reset your password," AI search recognises that these describe the same action and returns the relevant paragraph directly.

> [!NOTE]
>
> **Data privacy:** Nulix processes your article content to build and maintain the semantic search index. This content is **never used to train external AI models** or shared outside your workspace. Indexing happens within Nulix's secure infrastructure. For full details, see the Nulix Privacy Policy.

> [!TIP]
>
> Longer, well-structured articles with clear headings give AI search more context to work with and significantly improve accuracy. Break articles into logical sections with descriptive H2 and H3 headings rather than writing one long block of text.

## Next Steps

> **[Search Overview](https://docs.nulix.io/search/overview.md)**
>
> Review how standard search ranking works and learn how to use search analytics to close content gaps.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.