Intelligent Virtual Assistants

Definition of Intelligent Virtual Assistant (IVA)

An Intelligent Virtual Assistant (IVA) is software that performs tasks or services for an individual based on natural-language commands or questions. It is an application of Artificial Intelligence technologies, most prominently natural language processing (NLP) and learning.

The most popular platforms are (in descending order) Google Assistant, Apple Siri, Amazon Alexa, and Microsoft Cortana.

A lot of similar and related terms exist, often defined unclearly and used inconsistently. The most common use of some important terms is:

• Intelligent Personal Assistant (IPA) is a synonym for Intelligent Virtual Assistant (IVA).
• Chatbot is software that interacts with humans via textual or auditory means. While the boundary to IVA is floating, ‘Chatbot’ is typically used for systems that can only produce and understand a fixed, predefined set of words or statements, like guiding a user through phone menus.
• Intelligent Virtual Agent is an animated, artificial intelligence virtual character, usually with anthropomorphic appearance. Beyond natural speech, it also performs adequate non-verbal behavior like facial expressions and gesture. An example is Synthesia.
• Conversational Agent is either used as a synonym for IVA, or as an umbrella term for any type of software that converses with a human. This would also include a user interface that is for example based on gestures.
• Virtual Assistant is a job role carried out by a human, in which the tasks of a traditional, directly employed office assistant are outsourced to a service provider that offers remotely working ‘virtual’ assistants.

Market - Current Adoption

Relevance for IoT and integration

A) Consumer IoT: Convenience and fun

For consumers, the value of intelligent virtual assistants for controlling IoT objects is obvious. A large portion of private IoT is home automation, and for ‘smart speakers’, performing tasks at home is a primary use case. For these IoT products, supporting intelligent virtual assistants is a must. IVAs constitute an integrated user interface for many devices at once.

According to Amazon, in January 2020 there are more than 100,000 Alexa-compatible smart home products from more than 9,500 brands. The market for ‘smart speakers’ is still growing, but at least in the US, there are signs that it reaches saturation.

Another IoT-related consumer market worth noting is in-car IVAs. Due to the obvious advantages of hands-free operation, this became ubiquitous and is no longer an attribute of expensive cars.

B) Industrial IoT: Visualization and control of IoT devices

The relevance of IVAs for Industrial IoT (IIoT) is far less obvious. In those environments, typically a large number of IoT devices generate a stream of event data, which users need to be able to view, aggregate, filter, etc. The visualization with an interactive dashboard remains the best option to make large amounts of event data digestible.

When it comes to issuing commands, the priority of industrial applications is neither convenience nor a low learning curve, but correctness and precision. A crane operator will certainly still prefer to operate the crane with a sidestick. The imprecision and risk of misunderstanding that lies in the nature of natural language are even less acceptable with more at stake.

Further impediments of intelligent virtual assistant usage in professional environments include environmental noise and lack of continuous online connectivity.

Nevertheless, there are professional use cases. One is if the hands and eyes are occupied. The mentioned crane operator may want to control the heating of the cabin by voice command. Companies like Spectra offer industrial voice control solutions and corresponding offerings. For now, IVAs are a good additional option in IIoT, but can hardly be called disruptive.

C) Customer Service: verbal interaction on complex IoT related topics

IVA adoption is strongly growing in banking, telecommunication, healthcare, government, and all other sectors where customer service and interaction plays a major role. Using them in customer service is not an IoT use case per se. However, a lot of event data is generated in the context of Customer Service processes, so connections do exist. For example, in banking streaming analytics can be used to generate alerts if a customer’s stock portfolio develops a certain way. These alerts are not issued by a simple text message, but by the customer being called by an IVA. The IVA explains the situation and supports advanced questions like ‘how is my stock portfolio’s volatility this year compared to the NASDAQ?’

Market - Outlook

The overall strong growth of IVA usage will continue, even if specific products in specific markets (like smart speakers in the US) are nearing saturation.

This will include industrial use cases. However, a clear advantage of voice communication over other means of communication has to be established upfront; just being a novelty is not a sufficient driver for adoption in professional environments.

The Artificial intelligence that generates the content has been showing massive improvements, and ChatGPT's public release (beta) in November 2022 has been a breakthrough in terms of public attention. This is discussed in more detail in the Technology Radar entry on Generative AI. This will certainly be a strong contributor to the further growth of IVAs, the full impact of which is hardly foreseeable today.