Importing Files to Train Your Chatbot

Learn how to add PDF, Word, Excel, or text documents to build your DoxyChat assistant’s knowledge base.

Introduction

For your chatbot to accurately answer your users’ questions, you need to provide it with material. Importing files is the most direct method for transmitting your brochures, procedures, catalogs, or technical sheets to the artificial intelligence.

Supported Formats

DoxyChat supports the following standard document formats:

• PDF (.pdf): Ideal for brochures, user manuals, and official documents.
• Word (.docx): Perfect for internal procedures and administrative documents.
• Text (.txt): For raw content without formatting.
• Markdown (.md): Recommended for structured technical documentation.
• Excel (.xlsx) and CSV (.csv): For structured data (see dedicated section below).

Important note: Make sure your files are not password-protected, otherwise the AI will not be able to read them.

Specific Guide for Excel and CSV

Artificial intelligence does not “see” a table the way a human does (visual grid). For it to understand your data, DoxyChat uses a linearization technique: each row of the table is converted into a sentence comprehensible by the AI.

To guarantee quality responses, follow these best practices before importing an Excel or CSV file:

1. One mandatory header row: The first line of your file must contain the column titles (e.g., “Product Name,” “Price,” “Description”). The AI uses these titles to give meaning to the values.
2. One row = One object: Each row must represent a complete entity (a product, an employee, a branch). Avoid merged cells across multiple rows.
3. Clean the data: Remove empty or unnecessary columns to save your character quota.

Example of ideal structure:

The AI will read this as: “Product: Mountain Bike, Price: $299, Features: 27-inch wheels…”

Import Procedure

1. Go to your Dashboard.
2. Click on the Sources tab of your chatbot.
3. In the Add files area, click to browse your computer or drag and drop your documents.
4. The import starts immediately.

Statuses and Timeframes

Once the file is uploaded, it goes through several states:

• 🟠 Processing: The AI is reading, cleaning, and indexing the content. This can take from a few seconds to a minute depending on the size.
• 🟢 Complete: The document is ready. The chatbot can now use this information to respond.
• 🔴 Error: The file could not be read (corrupted file, unsupported format, or protected).

Limits and Quotas (Hard Limits)

Depending on your subscription plan (Discovery, Starter, Growth…), technical limits apply to guarantee service performance:

• Maximum size per file: From 5 MB (Discovery Plan) to 50 MB (Agency/Enterprise Plans).
• Number of characters: The AI analyzes the text, not the number of pages. A very dense document will count more than an airy document. Limits range from 100k characters to 4 Million characters per document.

If you exceed these limits, an error message will prompt you to split your file or upgrade to a higher plan.