---
title: "How to split a Fast Line Pro knowledge-base file with Ask AI"
description: "Semantic chunking via Instance Agent: Ask AI on the document segment settings screen."
---

# How to split a Fast Line Pro knowledge-base file with Ask AI?

If you do not want to tune separators and fragment size by hand, open document segment settings and click **Ask AI**. ConnectiveOne opens **Instance Agent**, attaches the uploaded file, and helps split the text into **meaning-based chunks** without rewriting the source — then saves them to the knowledge base.

Manual chunking ([configure chunking](./configure-chunking.md)) stays available — Ask AI complements it.

## When you need this

- The file is long and has no convenient separators (`---`), but you need splits by section meaning.
- You want to prepare a knowledge base faster without editing every segment by hand.
- You must keep the source wording (the agent must not paraphrase the text).

## Key terms

- **Ask AI** appears in segment settings when **Instance Agent** is available on the instance and is **not** in read-only mode.
- Uploaded document files are attached to the agent chat automatically.
- The agent splits text into self-contained semantic fragments and saves them through the usual segment flow.
- If the button is missing — contact your administrator (Instance Agent / permissions).
- This is **not** the same as **Ask AI** in the agent instructions editor (prompt writing help).

## Before you start

- [x] **Menu → Fast Line Pro → Knowledge base**.
- [x] You have a knowledge base and an uploaded file (or you are on the create-document step with a file).
- [x] Instance Agent is enabled and not read-only.

## Step-by-step

1. Open the knowledge base and start adding / configuring a document with a file.
2. Go to the **segment settings** (chunking) screen.
3. In the block titled like **Do not want to configure chunking manually?**, click **Ask AI**.
4. In Instance Agent chat, confirm the file is attached; optionally clarify the knowledge-base name or fragment size preferences.
5. Wait until the agent prepares and saves the segments.
6. Return to the knowledge base and review the document segment list.

## What happens next

The document has semantic segments ready for agent search. Next you can connect the knowledge base to an agent or fine-tune individual segments manually.

## How to verify

- The document shows several segments with source text (not a paraphrase).
- A test question to an agent using this knowledge base finds the right fragment.
- If Ask AI is unavailable, manual chunking still works.

## Related pages

- [Configure chunking](./configure-chunking.md)
- [Configure segments](./configure-segments.md)
- [Upload files to the knowledge base](./upload-files-to-kb.md)
- [Fast Line Pro integrator hub](../integrator-hub.md)