Frustrated with a simple use case for Fin AI

Hi ​@Heather Willson! I believe, this is something we discussed recently on the Community Office Hours session: basically, Fin has internal guard rails for the case where users want to escalate the convo to the agent that overwrites Guidance and proceeds with escalation. My assumption was that it’s done to avoid the infinite loop where users can’t escalate to support, but looping in ​@Roy and ​@Nathan Sudds here, just in case you folks know better :)

Personally, I would tackle this by inserting the Fin module into the workflow and defining escalation logic using the Workflows, where both Project and Organization can be asked directly. Branches checking for these attributes prior to the question would eliminate cases, where user is asked for this twice. 

Fin treats escalation as a fork rather than a sequence - it asks that question because it thinks that both paths are available at the same time.  Depends what you have tried but potentially guidance wording like “You do not have permission to escalate or handoff until you have collected both the Project name and Organization name from the user” could fix it, because prohibiting it is stronger than ordering.  But probably the fix is through workflow, an “ask a question step”, before escalation

“where both Project and Organization can be asked directly.” Directly, as in, manually asking for the user to pick or type their project? Anything manual like this is a non-starter for us and is why we’re using AI.

I am trying this out with a procedure instead, but having a hard time asking it to trigger on every conversation for a given audience, without exception. 

​@Nathan Sudds  would you happen to have a recording of the community office hours where this type of thing was discussed? Thanks!