Docx file

Use the Docx file option to convert Word documents (.docx) into EbbotGPT knowledge. You can upload a Word document directly or provide a URL to your file through the interface.

Conversion Process

The system uses Pandoc to convert the Word document into a Markdown file, which is then integrated into EbbotGPT as split documents. This makes the content searchable, usable for generating responses and displaying them as sources in the chat widget whenever EbbotGPT uses the content in its responses.

How it works

When EbbotGPT processes a Word document, it doesn't just "cut" the text; it attaches the current heading as metadata to every single chunk of information. You can think of this as a "sticky" label that follows the text until a specific event tells it to change.

1. The "Same-Level" Replacement Rule

A heading level remains active and is added to every subsequent chunk of text until it is explicitly replaced by a new heading of that exact same level.

If you have a Heading 2 titled "Shipping Policy," every chunk following it will be "tagged" with "Shipping Policy." This continues through multiple chunks until the AI encounters a new Heading 2 (e.g., "Return Policy"). At that moment, the "Shipping Policy" tag is dropped, and the "Return Policy" tag takes over.

2. Multi-Level Hierarchy (Headings 1–4)

Complexity isn't an issue because the logic scales across levels. EbbotGPT tracks levels 1, 2, 3, and 4 simultaneously. Each level operates on its own "replacement track":

• Heading 1 (The Parent) stays attached to all chunks until a new Heading 1 appears.

• Heading 2 (The Child) stays attached until a new Heading 2 appears.

• Headings 3, 4, 5 and 6 follow the same rule.

A single chunk of text deep in your document might carry the context of its H1, H2, H3, and H4 all at once. This ensures the AI always knows the full "path" of the information (e.g., Support > Hardware > Laptops > Battery Replacement).