It is now possible to add "exit keywords" in the AI - NLP rules by preceding them with an _ (underscore).
If you manage a chabot on cars with a context by brand. In the "Nerault" context you need rules like "breakdown", "maintenance" but also on the "Tiaf" brand to change the context if necessary.
If the user indicates in the context "Nerault": "otherwise I also have a Tiaf but it is broken", the expression "failure" may take over because it is certainly more popular in this context when it should be send the user to the context linked to the "Tiaf" constructor.
The solution so far: the negative keywords "failure -tiaf -geupeot -dorf -kosda -etc ...". A nightmare for the chatbot designer and this is to be expected for each expression in all the contexts of the brands or all the expressions composed on all the other manufacturers "Tiaf failure" "Kosda repair" "Dorf maintenance" .. Even an Excel formula will have a hard time generating all the combinations to import.
Now you just need to add just ONE rule containing simply: _tiaf
"Do you have any advice for a broken Tiaf?"
The Exit Keyword "_Tiaf" will take priority and immediately force an exit to the destination sequence which can change the context user on the way if necessary.
It is possible to add "exit keywords" at will and not necessarily in a context, it can also be in the general context and in this case the interest is to have words which necessarily take precedence over the others if they are in competition during processing.
Warning: No vehicle was abused during the development of this feature and any resemblance to existing or existing car manufacturers would be purely coincidental ;-)