Assembled AI Knowledge Management: How does it work? – Assembled

Assembled AI Knowledge Management

Assembled AI includes a robust series of knowledge base integrations so that agents can have all of the documents they need, at their fingertips.

To see a full list of which knowledge base integrations Assembled AI supports, click here.

Document tags

Document tags are an important part of how Assembled AI organizes your knowledge base. Tags are indicators that help Assembled AI better retrieve and parse articles when replying to agents.

Tags can be applied to a document in two ways:

  1. Automatically - when the document is synced, Assembled AI will automatically pick up pre-set fields from the document, and apply them as tags.
  2. Manually - on the Assembled AI admin view, you can go in and manually apply tags to articles as you see fit. We typically only recommend this as one-off workarounds.

To see how tagging works for your specific knowledge base, click here to read more about how Assembled AI integrates with your knowledge base.

Ticket field tagging

Similar to document tags, ticket field tags are also helpful to Assembled AI when retrieving documents. You can configure what ticket fields you'd want Assembled AI to use.

So far, we support the following ticket fields / properties:

Have a field in mind that you don't see? Feel free to reach out to a member of the Assembled team, and we can get that in for you right away!

Here's a common use case for Ticket + Document tags:

Note that if you do have ticket field tagging enabled, Assembled AI will only reference articles that are explicitly tagged by the ticket fields you pass in.

Ticket field context

Separately, you can also configure a set of ticket fields to pass to Assembled AI as context, for drafting a reply. A use case here could be you have specific tags associated with your customers, and you want Assembled AI to know which tags the requester is associated with when drafting a reply.

Assembled AI supports the same ticket fields for context as you see for tagging.

Image Reference: