Skip to main content

The Training Process

After adding content, you need to train your agent for it to actually learn. Here's how training works.

What is training?​

Training is the process where TAB takes all your content (documents, FAQs, text) and teaches it to your AI agent. Your agent processes this information so it can use it to answer customer questions.

When to train​

Train your agent:

  • After you first add content
  • After adding new content
  • After updating existing content
  • After removing outdated content

Important: Adding or changing content does nothing until you train. Your agent keeps using its old training until you run a new training session.

How to train​

  1. Click on your agent
  2. Click the Data Sources tab
  3. Review your content
  4. Click Re-Train (or Train during initial agent creation)
  5. You'll see a confirmation asking "Are you sure you want to train your agent?" with a count of your data sources
  6. Click Yes to confirm
  7. Wait for training to complete

What happens during training​

  1. Processing - TAB reads through all your content
  2. Learning - Your agent learns the information
  3. Indexing - Content is organized for quick retrieval
  4. Completion - Your agent is ready to use the new knowledge

How long does it take?​

Content AmountTypical Time
A few FAQs1-2 minutes
Moderate content3-5 minutes
Lots of content5-10 minutes

You'll see a progress indicator while training runs.

Training status​

Look for these indicators:

  • Not trained - You have content but haven't trained yet
  • Training - Currently processing
  • Trained - Ready to use
  • Failed - Something went wrong (see troubleshooting below)

Training and data storage​

Training uses your plan's data storage limit, not conversation credits. Each plan includes a set amount of storage for your training content. If you've reached your storage limit, you'll need to remove some content or upgrade your plan before adding more.

Managing your storage​

  • Remove outdated or unnecessary content to free up space
  • Keep content focused and relevant
  • Check your storage usage in Settings > Usage

After training completes​

  1. Test in the Playground - Ask questions to verify your agent learned correctly
  2. Check specific topics - Test the new content you added
  3. Verify accuracy - Make sure answers are correct

If something's wrong, update your content and train again.

Training failed?​

Data storage limit reached​

Check your data storage usage in Settings > Usage. You may need to remove some content or upgrade to a plan with more storage.

File issues​

If a document caused problems:

  • Check the file format is supported
  • Make sure it's under 10MB
  • Try re-uploading
  • Try a different format (convert DOCX to PDF, etc.)

Still not working?​

Remove content piece by piece to find the problem, or contact support.

Retraining​

When you need to update your agent's knowledge:

  1. Make your content changes
  2. Run training again
  3. Test the changes

Each training session builds fresh knowledge from all your current content. Old, removed content won't be remembered.

Tips for smooth training​

Review before training Make sure all content is accurate and necessary.

Don't train too often Batch updates together to save time.

Always test after Don't assume everything worked - verify in the Playground.

Keep records Know what you trained with, so you can troubleshoot issues.

Next steps​

After training: