Knowledge

How many knowledge pieces does GenAI use per question?

Currently, the Engine uses the top 3 chunks from your organization's knowledge. It prioritizes them based on a number of factors, such as textual similarity and similarity rank, through various search models. For more information, see this page.

Is there a way to prioritize certain knowledge sources rather than others?

There is no way to prioritize one source over the other. You can limit the amount of unnecessary content in your environment. For more information, see this page.

Can knowledge sources contain tables?

Yes. See this page for more information.

Can knowledge sources contain images?

Images are currently not supported. See this page for more information.

Can knowledge sources contain (bullet) lists?

Yes. See this page for more information.

Can a knowledge source consist of multiple languages?

Yes, but it is not recommended. See this page for more information.

How long can a synchronization take?

The crawler can crawl for up to 48 hours in one go. If the crawler synchronization isn't done within 48 hours, please review your added web pages (especially domains that use a wildcard). See this page for more information.

I just started using GenAI. Can it go through my archived existing conversations (MSC)?

No, collective memory only starts running post-onboarding, meaning we learn from new conversations provided there is enough to go on. Even then we don't look back further than 2 months, because information expires. It's not sensible to use data from 3-year-old conversations, for example, because that will probably hold no longer true or accurate information.

---

Behavior

How can I set the right tone of voice?

You can find the do's & don'ts for setting the tone of voice on this page.

---