This idea is similar to the feature request for “CLAudia having context from previous recent conversations from the same customer” (https://cloudhumans.canny.io/feature-requests-claudia/p/4573-claudia-ter-contexto-de-conversas-anteriores-recentes-do-mesmo-cliente). But in this case, instead of looking at past conversations, the focus would be on the attributes (metadata) of the contact.

In CloudChat (as well as Zendesk, Intercom, etc.), contacts generally have multiple attributes, such as:

Today we already ingested this data from the helpdesks via integration. However, CLAudia still does not receive this information during the service. This makes her have to ask for data that, in theory, should already be available to her. For example, asking for the email only to then, via Eddie, pull the contact's metadata from an external API (or from the helpdesk again via Eddie as well).

Even when Eddie is used and consults all the data, CLAudia only has access to what was actually mentioned in the response to the customer. Example:

“Hello Fulano, I confirmed that you have 2 requests: 123 and 456!”

In this case, CLAudia learns about these requests because they were publicly mentioned in the conversation. But if, during the same consultation, Eddie also brought “profile” or “pending payments”, these data are not accessible to CLAudia if they are not mentioned in the answer. If the customer then asks “How many payments do I have pending?” , it would be necessary to activate Eddie again, triggering a new API call and facing the same limitations.

This feature would solve exactly that point: allowing CLAudia to have, from the beginning of the service, the full context of the contact's attributes, without depending on reinquiries or what was said in the conversation. This would amplify RAG's powers and we could create more specific Eddies for corner cases based on that metadata.