Manju is a lead software engineer at Salesforce, where she works on Einstein Voice Assistant, bringing the best in class artificial intelligence capabilities to Salesforce CRM. Prior to Salesforce, she worked in a breadth of roles solving complex problems at Google, Goldman Sachs, NVIDIA and Akamai. She holds a Masters in Computer Science from Texas A&M University. She is a new mama to a baby girl and loves outdoor adventures, reading and blogging in her free time (joke! there is no free time as a new mom).
What if using a business application was as natural as having a conversation? Could a Voice Assistant do more than just tell about the weather or play songs? Come learn a practical framework for building an AI Voice Assistant for Enterprise, that completes business tasks through voice! Using Voice and Natural Language Understanding (NLU) we translate unstructured data to structured inputs in the system, while enhancing the end user experience. In this talk we will go through the high-level architecture and workflow of the Assistant. This starts with Automatic Speech Recognition (ASR) on device to using NLU for identifying entities and intents in a single dialog conversation text. And translating intents to structured inputs in database. We will dive deeper on specific components of the architecture like the APIs and the core language services that power the Assistant.
Come to learn our practical approach to implementing a Voice Assistant and the unique challenges involved in interacting with data from complex sources. We will share lessons learnt building the first few iterations of the Assistant and discuss further opportunities to expand on our approach. All the language services are open source and available on github. Key takeaways for audience:
As NLU capabilities get more accurate these can be translated to structured inputs providing additional business insights. 4) Prepare the audience to think long term about Voice as the next User Interface to interacting with data