I have the following trigger phrases:
And I have the following entities:
I was hoping this phrase would either start the dialog because it contains the trigger, or - slightly more complex but it was worth testing - see if the agent is smart enough to realise that an entity is in the same sentence as the trigger phrase and extract it along with the trigger to move along the dialog to the point where the entity is required.
Instead I get a fail on either expectation 😛
Here is a more complete view of the dialog.
As mentioned I'd hope in this case either the trigger would work, or even better the trigger would work AND the entity would be detected in the initial phrase and move the dialog right along to the point where the entity is used. Otherwise you have to impose a more artificial conversation flow where it's more back and forth to establish entities. Often, people already know the entity they want information on, so they will concatenate the request. E.g. you don't naturally ask someone "Can I have the phone number of someone?" then they respond "Sure, whose phone number would you like?". You tend to just say "Can I have James' phone number?".
Solved! Go to Solution.
Hi @JamesM ,
Your idea is really good, but the chatbot is a fixed or more fixed conversation that AI bot. AI bots read phrases/words and get the action related to phrase or word using artificial intelligence, but the chatbot (PVA) is more simple than AI bots.
You can try to use artificial intelligence and knowledge base or something like that on Power Automate, but unfortunately, you will not do it in the PVA yet.
Hi @JamesM ,
Your idea is really good, but the chatbot is a fixed or more fixed conversation that AI bot. AI bots read phrases/words and get the action related to phrase or word using artificial intelligence, but the chatbot (PVA) is more simple than AI bots.
You can try to use artificial intelligence and knowledge base or something like that on Power Automate, but unfortunately, you will not do it in the PVA yet.
Hi All,
We are currently using the Power VA trial and are discovering the same limitations with the trigger phrase. We are also trialling IBM's Watson Assistant and in comparison the AI seems to handle these much better. Power VA seems to require much more specific requirements to trigger the correct response.
From a platform perspective Microsoft Power VA seems to be a much a better fit for us with the integration, deployment options and ease of providing authentication. However, it seems that it will require much more work inputting the correct trigger phrases to produce a required response.
Does anybody know if there are plans to improve these capabilities within the Power VA platform to make a response more intuitive for variations in the trigger phrases / utterances??
Keep up to date with current events and community announcements in the Power Virtual Agents community.
A great place where you can stay up to date with community calls and interact with the speakers.