Library & AI Training

How to train Spara AI using the Library.

Spara AI is trained on three data sources:

Webpages

Spara AI is trained on your publicly available marketing website. This can include multiple subdomains, for example including your product documentation site. This is where Spara AI learns the bulk of information about your company.

You can see which webpages are included in your training data set on the Webpages tab of your Library page.

Media

Spara is trained on the text included in PDFs, images, videos, and other files that you upload to Spara.

You can see which uploaded media files are included in your training data set on the Media tab of your Library page.

Contact your customer success representative to upload media or to include/exclude certain webpages from your training data set.

Key Questions

Key Questions represents the most important information for Spara to answer accurately. These questions can be about anything: value prop, important product features, your sales process, etc.

Each Key Question has a Current Response and an Ideal Response. The Ideal Response is inputted by you - it represents the best possible answer Spara could give to this question. Each time you edit an Ideal Response, you will see Spara recalculate it's Current Response to more closely match your Ideal Response!

Don't expect Spara to copy Current Responses word-for-word; Spara AI is designed to incorporate all the important information into its responses.

Key Questions capture the most important questions your leads will ask Spara.

Feedback given in Spara's test interface is also incorporated into Spara AI learning over time (see section below)