How much information can I use to train an Interactive Avatar with extensive company data?

Hi HeyGen community,
I'm looking to create an Interactive Avatar that can answer detailed questions about my company. I have a large volume of content I’d like to include—multiple web pages, internal documents, presentations, and more.

My main question is:
Is there a recommended or technical limit to how much content (in terms of words, document count, file size, etc.) I can use to train or feed the avatar?

Additionally, I’d appreciate any advice on:

What data formats work best for this use case?

Are there any best practices for organizing large-scale content for optimal performance?

Thanks in advance for any guidance!