At Five9 we believe that Contact Centre AI needs to be engine-agnostic, embedded, and practical to deliver meaningful business value.
One way that we’re delivering on this promise is by embedding access to large language models across our AI portfolio. Our AI Summaries and AI Insights products are now powered by OpenAI’s LLMs and at this year’s Five9 CX Summit, we announced support for OpenAI in Five9 IVA Studio.
In keeping with our strategy of remaining engine agnostic, support for OpenAI now provides developers with another set of Conversational AI tools in addition to our portfolio of Speech-to-Text, Text-to-Speech, Sentiment Analysis, and Voice Biometrics engines.
We also announced support for Google’s automatic language detection and for Deepgram’s Nova Speech-to-Text engine. You can watch the full product keynote here.
We believe that these enhancements will make it easier and more practical for our customers to build conversational applications that improve the self-service experience for consumers.
Intent Detection using LLMs
Five9 IVA Studio now allows task designers to use OpenAI to determine the caller’s intent. This serves as an alternative to traditional NLP engines and eliminates the need for training phrases to optimize intent matching.
Datastore Driven FAQs
One of the most popular uses for IVAs is to answer frequently asked questions. These FAQs can now be stored in a Studio Datastore and maintained using csv upload or using datastore APIs. Generative AI can then be used to understand which question was asked and return a result from the Datastore.
Generative AI can now be used to summarize the conversation between the IVA and the customer. That summary can then be passed over to a live agent when the call is escalated and/or be pushed into a CRM for future reference.
Custom LLM Prompts
Studio task designers can now build their custom LLM prompts giving them the freedom to use Generative AI in a wide variety of ways. For example, they could simply ask OpenAI to extract a package number or policy number from caller utterance (e.g., I’d like to get the status of policy number 76788TYF”)
LLM Code Crafter
Try it out yourself.
Call (313) 552-4693 to try a demonstration of a self-service banking application that uses OpenAI to determine the caller’s intent (e.g., “I want to check my balance”), answer frequently asked questions (e.g., “What’s a guarantor”) and to summarize the call at the end of the conversation (that summary will be sent to you by SMS).
Automatic Language Detection
Five9 IVA can now understand which language a caller is speaking. It can identify up to four languages by listening to the caller speak. This feature is now available in Five9 IVA Studio through our Cloud Speech-To-Text Node for Google.
Improved Speech-To-Text Accuracy with Deepgram
One of the biggest challenges organizations face when developing speech-driven self-service applications is numeric and alphanumeric speech-to-text accuracy.
For example, an insurance company might need to capture the caller’s policy number, or a shipping company might want to get the customer’s package tracking number before they can offer self-service. So, better accuracy naturally translated into higher self-service containment rates which in turn drove more cost savings.
Five9 IVA Studio now provides access to Deepgram’s highly accurate speech recognition engine through our Cloud Speech-To-Text node.
To try it yourself call 219-695-8627. You’ll be able to speak a ten-digit set of numbers and/or letters and the IVA will repeat them back to you to confirm that it understood you correctly.This blog post has been re-published by kind permission of Five9 – View the Original Article
For more information about Five9 - visit the Five9 Website
Call Centre Helper is not responsible for the content of these guest blog posts. The opinions expressed in this article are those of the author, and do not necessarily reflect those of Call Centre Helper.