Virtual assistants don’t need developers or AI | Tech News
Virtual assistants can go beyond playing music or buying things on Amazon. And they don’t need corps of developers or fancy artificial intelligence technology to do so.
Business and society in general have moved a long way towards self-service. Most offices no longer have a secretarial pool, and most managers and even some executives don’t have a dedicated assistant. Travel agents and other common jobs of yesteryear still exist but you don’t go to a travel agent to book a simple direct domestic flight; you go to a website.
Gartner predicts that 25 percent of companies will use virtual assistants in customer service in the next couple of years. Google recently demonstrated a virtual assistant that makes phone calls to book appointments.
However, self-service isn’t always more productive. If you have 40 people booking flights to a company meeting, you’re probably better off using a travel agent. What about when you have multiple SaaS applications necessary to complete a task? Moreover, why should a busy executive have to learn three software applications just to get an answer?
An enterprise virtual assistant (EVA) could answer a question like “What is my sales quota?” An EVA could also perform tasks like “Book me a flight for Thursday’s QBR meeting” or “Find me a meeting room and invite Susan Smith for next Thursday to go over the technology marketing campaign.”
Under the covers, the EVA is logging into salesforce and doing a search, hitting the corporate travel site and the calendar system. But the executive doesn’t need to know this, at least not when asking a simple question.
Much of this isn’t machine learning or deep learning. Much of a virtual assistant is simple rules and commands. In fact, an EVA is largely a voice interface connected to a command-line interface and to essentially some devops scripts. This isn’t to say it won’t use AI, but that its basic capabilities don’t.
Current virtual assistants like Alexa and Siri are largely focused on the individual and can only perform relatively simple tasks. They can be extended with skill kits that are largely dependent on developers, but actually most companies are already instrumented for what a virtual assistant needs. Not everything a person does has a REST API, but most modern corporate apps have REST APIs. Some of those APIs are even decent.
An EVA can also learn new tasks without a developer. There isn’t a good reason why a combination of strategies can’t allow an EVA to learn to use a new API from voice commands and perform a new task. This is just a more advanced way that chatbots learn to answer new questions. This removes the need for a developer to get involved just because some integration is required.
While an EVA is really a smarter command-line interface, EVAs may bring about the full loop in human development. People went from talking to writing to typing to using the mouse to using a touchscreen. With an EVA, we may all return to a more conversational, less screen-oriented future.