Google's conversation AI platform Chatbase was built at Area 120, the tech giant's workshop for experimental efforts that may or may not succeed. Chatbase was first brought to the public eye at I/O 2017 as a chatbot analytics platform. This service, as only part of the whole product, was later rebranded as Chatbase Virtual Agent Analytics.
Today, the company has announced Virtual Agent Modeling as a new service that is part of Chatbase, to help businesses design more versatile virtual agents through a much quicker process. The new offering is currently available in the form of an early access demo for interested parties.
With development of Chatbase having continued over the past year, Google believes that customer service virtual agents are the best way in which its product can be deployed. Furthermore, the firm has also leveraged machine learning to analyze live-chat transcripts, deeming it as another core aspect in the course of the development of this technology.
Virtual Agent Modeling is said to reduce development times of a voice or chat virtual agent by a factor of 10. Developers will also be able to export results of the undertaken analysis on the data - provided through interactions - to virtual agents such as DialogFlow. Additionally, the modeled agent will be able to handle 99% of interactions, with accurate recognition of moments where calls that are outside its scope need to be transferred to a live agent.
Currently, Chatbase Virtual Agent Modeling is only being offered in the form of a personal demo. In order to qualify for a demo of the product, at least 100,000 English-language live-chat transcripts will need to be made available by the interested company. More details regarding the Virtual Agent Modeling can be obtained here.
If all turns out as planned, Google's AI-powered virtual agent could radically improve customer service experiences. However, it is perhaps too early to ascertain anything at this point.