ENDS SOON: 50% OFF FLASH SALE!

d
h
m
s

AI Training

Overview

The AI Training section is used to manage all training data uploaded to the Retrieval-Augmented Generation (RAG) system. This area allows administrators to upload files, import URLs, assign categories, manage expiration settings, and monitor indexing status for AI training items.

The system supports multiple types of training sources including files, URLs, support tickets, and BetterDocs content.

Anyone who has capability ‘Manage AI Training’ can:

  • Add new training items
  • Upload PDF and TXT files
  • Train AI using URLs
  • Assign categories to training items
  • Set expiration periods
  • Monitor indexing status
  • Delete training data from the AI provider and database

Each training item is processed and indexed into the RAG system for AI retrieval and response generation.

 
Where to Find
Support → Settings → AI Assistant → AI Training

How It Works

When a new training item is added:

  1. The item is queued for processing.
  2. The training data starts uploading to the RAG system using a cron event.
  3. The system processes each item one by one.
  4. After successful indexing, the status changes to Indexed.
  5. If processing fails, the status changes to Failed and the system automatically removes the failed training data.

Training data can originate from:

  • Uploaded files
  • URLs
  • Support tickets
  • BetterDocs articles

 

Supported Training Sources

The AI Training system supports the following sources:

  • File: Upload TXT or PDF files
  • URL: Train AI using website content
  • Ticket: Train AI using support ticket data
  • BetterDocs: Import BetterDocs documentation

 

Expiry System

Each training item can optionally have an expiry period.

Supported expiry durations:

  • Days
  • Months
  • Years

Default behavior:

  • Training items never expire unless configured.

When the expiry date is reached:

  • The training data is automatically deleted from:
    • The AI provider
    • The database

This helps maintain clean and up-to-date AI datasets.

 

Add New AI Training Item

  1. Navigate to:

    Support → Settings → AI Assistant → AI Training

  2. Click the Add New button at the top of the page.

  3. Choose the type of training source:

    • File Upload
    • URL
    • Other supported integrations
  4. Depending on the selected source:

    • Upload PDF/TXT files
    • Enter a URL
    • Select an integrated content source
  5. Select a category for the training item.

    • Customer Support Tickets
    • Technical Documentation
    • Product Information
    • Business Documents
  6. Choose whether the training item should expire.

    • Never expires
    • Expire after days/months/years
  7. After saving:

    • Status is initially set to New
    • The indexing process starts automatically by scheduled cron event

 

Training Status Types

  • New: The training item has been added and is waiting for processing or indexing.
  • Processing: The training item started to upload to selected provider.
  • Indexed: The training data has been successfully uploaded and indexed into the RAG system.
  • Failed: The training process failed or error occurred while uploading to AI provider.
  • Deleted: The training process failed will be deleted by cron event.

 

Bulk Actions

Bulk actions allow administrators to manage multiple training items at once.

  • Bulk Delete: Delete multiple AI training items in a single action.

 

Search and Category Filter

The AI Training page includes:

  • Search functionality
  • Category filtering dropdown

These tools help quickly locate training items.

 

Notes / Tips

  • Organize training data using categories for better management.
  • Use meaningful category names for easier filtering.
  • Expiry settings help automatically clean outdated AI training data.
  • Large datasets may take time to process and index.
  • Failed training items are automatically cleaned up by the system.
  • Review training status regularly to ensure successful indexing.
  • Deleting a training item permanently removes it from the RAG system and database.
What are your feelings