Genesys Cloud - Main

Thanks for reply Guru.

I have seen and read those links already.

Back to my point, there is still a lack of technical information and working examples from a flow perspective.

I have tried the AI Asks and they are still quite poor at the tasks mentioned from my own experimentation. The address capture (trying UK) is quite bad, a poor user experience when from a Voice source.

Does not help that the Interaction replay does NOT populate slots from an Intent capture, only an Ask object.

Hello Simon - Hope you are well and thanks for watching the video and sharing the comments. On 2nd Dec, there was another feature release which addresses your questions and feedback

Capture slot values via LLMs with Genesys Virtual Agent

Administrators can now enable virtual agents to collect slot data more conversationally using advanced natural language understanding (NLU) techniques powered by large language models (LLMs). This feature allows virtual agents to handle a variety of slot types such as free form, alphanumeric, and numeric; capture slots across multiple conversational turns; and respond to nuanced inputs like customers spelling out their names. This enhancement helps make virtual agents more interactive and closer to human-like conversations. For virtual agent authors, it simplifies the process by requiring them to specify the desired slots while the system manages complex interactions, such as completing partially provided information, handling branching dialogue, and addressing loops

These links may be very useful to explore this feature. 

https://help.mypurecloud.com/articles/virtual-agent-slot-authoring-recommendations-and-limitations/

https://help.mypurecloud.com/articles/use-ai-powered-slot-types-in-a-bot-flow/

https://help.mypurecloud.com/articles/free-form-slot-capture-examples/

Please take a look. Thank you.

I am starting out looking at the VA options and not seeing anything like real world example flows that shows how these intents and slots are mapped and used.

I can see the benefit of the AI generating intents from a description but that is an upfront developer time benefit. Not conversation real time.

I see conflicting recommendations, some saying keep intents clear and do not overlap purpose, whereas the video (and other limited documentation on this) shows a very generic all purpose description being used.

Are there any decent flow examples and Architect information on how this all works at a technical level, and valid intent capture with multiple slots.

I am experimenting with BOT flows at present but feels like this type of literature should exist to ease the learning path for users.

Feels all like marketing material and hype, and little in the way of technical detail and examples at present.

I am specifically interested in Voice, so Speech to Text capture and processing.

Thanks