Hi Markus,
your use case might sound quite simple in a first place, however from a technical perspective it is not. From what I understand is, that you want your bot to handle the second user input (after first has identified the user needs help) to be treated as kind of free text input. As this could be anything, this is nearly impossible to understand and label correctly by a natural language understanding engine. Therefore this requires special treatment - which is unfortunately not there yet, but planned for the near/mid term future of CAI.
There is still a way to overcome this situation, but it requires some effort and is not straight forward. Let me quickly sketch what you would need to do:
There is a first intent "problem" and a skill "identify-problem". The skill needs an intent based trigger "problem" is present (like you probably already have it) and then an action (not a requirement) with a message like "describe your problem", a set memory action where you set "problem-identified" as "true" and a goto action set to "wait for user input" and pointing to skill "handle-problem". This second skill has a trigger "problem-identified" is "true" and then performs the action you want to do - you will find the free text input from the user in the _source. Don't forget to set memory "problem-identified" to "false".
What happens is, that after the first skill was executed the system gives preference for the second user input to the skill "handle-problem" and with the memory based trigger, executes that skill.
Names/Labels are just chosen to give a more concrete example - of course you might choose them as you like.
Hope this helps and best regards,
Stephan
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.