Le Journal du Net ( JDN – 08 December 2020 ) looks at the choices available to the industry to select a suitable voice assistant. Voice assistants for the general public, voice assistants dedicated to specific functions… the offer becomes plethoric and complex to follow without applying certain strong discriminating criteria, specific to the industry.

SPIX industry offers a cross-section of the subject with elements drawn from its industrial and technological experience. The approaches of the digital newspaper and the supplier of industrial voice assistants serving the men and women of the industry can be complementary in answering the question of the price of free!

Voice Assistants: who are the real market players?

If we look at the number of voice assistants installed, all terminals combined, in 2019 we exceed the number of 2.5 billion. On this subject, the main players in the market do not hesitate to engage in a small battle of figures. When Amazon boasts of having sold more than 100 million copies of its speakers with Alexa, Google responds by announcing the installation of Google Assistant on more than a billion terminals by 2020 (c Célia Garcia-Montero, JND 2020).

Today, the major players in voice assistance known to the general public are GAFAM: Google, Amazon, Facebook, Apple and Microsoft. These American players are essential due to their power of technical development and their economic power. It is important to note that in this category of actor, no European can really be considered as a competitor.

By dint of talking only about the GAFAMs, we would almost forget the BATX which are quickly arriving on our markets, Baidu, Alibaba, Tencen, and Xiaomi which are coming to us directly from China. For its part, Samsung’s voice assistant is making a remarkable breakthrough in Asia and the United States. By way of comparison, in 2020 Chinese voice assistants are more deployed in the world than Apple’s “Siri”, as is Samsung’s Bixby.

This observation raises a first question about the sovereignty of European industry, as well as something to think about our models of economic and technological watch… Looking only at the tree, we risk missing the forest.

Can we do everything from a single economic model?

This frenzy for voice assistants is not without risk for users. Since their appearance, they have been the subject of multiple charges. They are suspected, among other things, of not fully securing conversations and keeping users’ personal data, which represents a serious privacy risk (c Célia Garcia-Montero, JDN 2020).

In this case, the user would like to have butter and butter’s money! The butter is to be able to have an overperforming voice assistant almost free of charge. Butter money is keeping control of your private life. Both GAFAM and BATX have decided otherwise for us…

The technical development of voice assistants as we know them today has cost a small fortune to each of the companies that develop them. These investments, which can only be made by the big players mentioned above, must be profitable one day. Two solutions are available to them: make the customer pay for a product to provide access to a service, or provide access to a product to monetize a service. It was the second choice that took over! The major voice assistant publishers offer us their products, so that we can consume services offered by other e-commerce or advertising players. The icing on the cake, in the general conditions of use of these “free” products, we all agree to pool our data and share our time spent listening to high value-added advertising content.

It is on the basis of this type of model that Amazon can train its voice assistant Alexa with several million data and exchanges between users, shared free of charge by all users of the product. This investment is then made profitable by the increase in sales through voice interfaces on the Amazon platform, and by the advertising content broadcast by other advertisers.

So that’s the price of free! The question to ask for the development of voice assistance solutions for the industry: is this price acceptable? If not, what economic model to adopt?

---

The monetization of data and the industrial constraints

To perform, Alexa has been trained with billions of conversations around the world. On the other hand, in industry, we do not have sufficient semantic data to constitute big data. Thus, BtoB voice assistants are not trained in the same way (c Célia Garcia-Montero, JDN 2020).

If, as we have seen, the economic model of the major voice assistants is based on the valuation of the data handled, then the question must be asked of the adequacy of this model with the capacities and constraints of the industry.

Unless the world of tomorrow is really different, it is unlikely that sometimes competing manufacturers will agree to pool all their documentary, semantic and linguistic resources in order to train a voice assistant. Even in a utopian world, the sharing of such data could possibly be envisaged at a precise moment, but it is not enough. The voice assistants of our giants need to be constantly supplied with fresh and qualified data in order to continuously increase their capacities.

It is illusory to think that manufacturers will make all the exchanges between their employees and a voice assistant available in real time to a third-party company, and authorize it to use this data for the benefit of the whole of the industry.

So, what do we do ? We could ask digital players on the west coast of the United States to sell a license for private use of this type of assistant for the benefit of a single industrialist. This is a hypothesis, but it will be costly, even unaffordable for the majority of manufacturers. Then there will be the question of the specific training of these assistants for the specific needs of this manufacturer. Why not, but we will quickly come up against the volume of available and accessible data. Let’s take a practical example: an industrialist who manages 100,000 different types of tasks for his maintenance operations is already a good size. Nevertheless, 100,000 documents are not much compared to the billions of exchanges used by Alexa to train.

In conclusion, whatever the technical solution envisaged, the manufacturer will find it difficult to accept the constraints imposed by the “free” models of the major players in the field. So you have to be creative and find other industry compatible approaches.

What is the solution for the industry?

A voice assistant, also called a smart personal assistant or connected speaker, is a device based on natural language voice recognition to allow its user to perform a search by voice (c Célia Garcia-Montero, JDN 2020).

To meet the needs of the industry in terms of voice assistance, it is necessary on the one hand to broaden the definition of a voice assistant, then to adapt the technology and the economic model to the specific constraints of this activity.

Voice assistance for industry will take different forms depending on the intended use. Indeed, an assistant for answering customer questions (FAQ type) will not have the same attributes as a voice assistant for generating field reports for site inspectors in the nuclear industry.

Two needs coexist for an industrial voice assistant, responding to a request from a user on the one hand, collecting and structuring information provided by a user on the other. Thus, for the industrial field, the definition of a voice assistant can be extended to “ smooth relations between an industrial employee and a sometimes complex digital system thanks to voice and intelligent assistance ”.

From a technical point of view, an industrial voice assistant will have to meet the specific constraints of the field: little data, a requirement for reliability, functional in noise, efficient offline, acceptable as a work tool by the employees of the industry. Still technically, an industrial voice assistant must be able to dialogue with all the content already validated and used operationally in industry. One can easily imagine that no industrialist will question what he has validated in SAP to set up a voice assistant… Therefore, any intelligent industrial voice assistant must take as input data the contents of digital tools already deployed like SAP, IBM-Maximo, Delmia-Apriso, INFOR, IFS, and so many other digital solutions.

The ” Spix ” industrial voice assistant from SPIX industry responds to these industrial requirements to better control production tools or systems under maintenance. Indeed, the investment to be made in the implementation of an industrial voice assistant is profitable if the manufacturer gains points of productivity or profitability of its operations. The voice assistant at the service of employees allows the manufacturer to collect more data on the state of his production tool, or to be better informed of the state of a complex system for which his employees are responsible. SIMSOFT INDUSTRY’s “Spix” voice assistant is at the service of employees’ expertise and allows them to more easily trace information from the field to the company’s information systems.

Finally, as the industry does not wish (or cannot) share its data to access “free” services, different economic models must be found. The major players in industrial digital technology such as SAP, IBM-Maximo, Delmia-Apriso, INFOR, IFS and others have already worked on economic models adapted to the expectations of their customers. The industrial voice assistant needs to find a compatible business model of these. For example, SPIX industry’s industrial voice assistant adopts the economic codes expected by its customers who already use industrial software.

---

Conclusion

Manufacturers want to benefit from the advantages of intelligent voice assistance technologies to help their employees (or their customers) in carrying out their tasks. In this professional world where data protection is not an option, it is urgent to bring out alternative European solutions to the American and Chinese giants. SPIX industry with its industrial voice assistant ” Spix ” is a pioneer in Europe in this field and is in a good position to take the leadership of this emerging market.

---

Press contacts
André JOLY – Managing Director
Phone. : +33 (0)6 25 17 27 94
Email: andre.joly@spix-industry.com

Legal entity
Website : spix-industry.com
Linkedin : linkedin.com/company/spix-industry
Simsoft3D SAS – 1244 rue l’Occitane – 31670 Labège (France)
“SPIX” and “SPIX industry” are registered trademarks of Simsoft3D SAS.