The digital transition is accelerating rapidly due to new personal and professional lifestyles brought about by the current pandemic situation. Conversational assistants are part of this transformation by enabling the automation of support and self-service requests.

The need for business-process automation continues to grow, driven by the health situation, which unfortunately is likely to continue for a few more months if not years. The year 2020 was a year of questioning, the year 2021 was a year of transition between a before and an after. The fragmentation of teams, due to remote working, imposes a digital transition for companies and the accessibility to their functions wherever they are and at any time.
Support functions are over-subscribed and the need to empower teams to do things for themselves is more than critical. Called self-services or shift-left depending on whether you are a customer or a solution provider, this approach consists of empowering customers, suppliers, and employees to get answers, solve problems or perform certain operations by themselves, 24/7.
In the category of automation means, we find conversational assistants powered by Artificial Intelligence. These allow to automatically handle several use cases such as: answering recurrent or occasional questions, executing more or less simple operations, helping to solve general computer problems, assisting humans in certain operations, etc.
Prediction 1: Demand Will Continue To Grow
The demand for Business Process Automation (BPA) is on the rise, driven by the need for companies, and more particularly business units, to find ways to save resources and deliver their services despite the global situation. Some support solutions are mature from the point of view of language processing but also from the point of view of the tools needed to make it work.
This maturity increases the success rate of these projects. The gains are even more interesting if they are increasingly complemented by other automation solutions such as Intelligent Document Processing (IDP), which enables documents sent by users to be processed automatically (invoices, order forms, ID cards, CVs, etc.), in real-time and with the possibility of interacting with the user on the data collected; and Robotic Process Automation (RPA) scripts, which enable access to systems without having to develop new connectors.
In 2022, market growth in Europe (excluding the UK) is expected to be 1.5 to 2 times what we experienced in 2021.
Prediction 2: Scaling Up
Some companies have already experimented with conversational assistance solutions, particularly in the consumer and professional services sector (BFSI, telecom operators, certain public administrations), where several conversational instances covering different domains have been deployed. In the retail/marketing sector, there is often a single instance responsible for answering buyers’ questions on the site. Bots are often accessed via a text-based interface, more rarely by voice. Organizations that manage many day-to-day relationships are of course the most receptive to these technologies.
The year 2022 should see an increase in usage with more specific cases and therefore a multiplication of deployments in those companies that have already launched projects and have acquired a good vision of the benefits and disadvantages of these solutions. The areas covered in priority: IT support (general requests, problem resolution, ordering equipment, etc.), HR (general questions, recruitment, new arrivals, etc.), sales (customer services), certain specific business areas (operations on my deposit account, complaints, etc.).
Companies that have not yet experimented with these technologies will have to think about the implementation of solutions. A large number of solutions on the market will not make it easy for them to choose, especially since there is usually a premium for the first salesperson to propose a solution, which can lead to negative and disappointing experiences.
Prediction 3: The Emergence Of The Master-Bots
With the proliferation of instances and the increase in the different business domains covered, accessibility to the different Bots will become problematic and will require the grouping of the different entry points in a single interface. This general assistant, also called Master-Bot, will help the user to solve multiple problems or tasks that require access to different business domains.

The Master-Bot is the single point of entry to the various services. It is the control tower for exchanges between the user and the business areas. This device allows the user to avoid having to search for the required expertise to answer his questions in various places. The domains are contacted by the Master-Bot according to the request.
Several solutions on the market have already implemented this mechanism and the first uses will come during 2022.
Prediction 4: Increased Testing With Voice
The general public is familiar with voice assistant devices like Amazon Echo or Google Home. However, these devices are not suitable for use in professional environments. Nevertheless, the uses of voice for professional purposes are multiple and relevant: as a complement to standard customer phone support (Callbot) or as a hands-free assistant in a situation (Voicebot) such as in a workshop, a laboratory, or a vehicle.
The technologies required to deploy a voice assistant are more complex than for a classic "Chabot", which makes these solutions more complex to master and projects more difficult to succeed.
In 2022, companies will continue to test voice to learn how to use this channel and the associated technologies (telephony, voice recognition, voice synthesis, voice dialogue management…).
Prediction 5: The Disillusionment Of Connected Assistants
The analysis of the requests made in Google (via Google Trend), on several names of connected speakers in the last 4 years, shows an overall decrease of requests in the United States.

These speakers suffer from several problems that slow down their use, such as the lack of trust in the use of personal data, opaque or too general services (unless you are a music lover or want to order products on Amazon), and the low level of home automation equipment in the home. They are seen as gadgets to be given at Christmas and not as vital equipment in the home.
This situation is expected to continue in 2022.
Prediction 6: The Decline Of Newcomers
Advances in automatic language processing and the availability of statistical models have made the creation of an assistant within everyone’s reach. Nevertheless, an assistant is not only a language processing unit but also: tools for modeling dialogues, piloting tools allowing the business to supervise the operation by themselves, connectors to users and information systems, conversation histories, advanced role management, behavior analysis units, etc. Most of the solutions on the market have several years of experience and development behind them, which creates a high barrier to entry for newcomers. They will need to have a breakthrough innovation that other solutions do not yet have, which will become increasingly difficult to find.
Moreover, the success of projects is not only based on technology but also on the mastery of project cycles, leading to a strong professionalization of the market with the need for support throughout the project and beyond. Especially since companies are more and more attentive to the return on investment of these solutions. Companies will have to think twice before starting the creation of a specific solution. Especially since some free Open Source solutions allow to realize Chatbots with lower license costs (but with higher project phases and maintenance costs).
The only FAQ-type solutions will not succeed in surviving in this ecosystem except by offering extremely low prices and/or by having a large base, which will nevertheless lead to a problem of long-term profitability.
The year 2022 should therefore see few new entrants and the disappearance of the most basic solutions. The current state of AI technologies should not allow a major technological breakthrough to emerge quickly (at least for industrial use). During the year, we will certainly see new consolidations, especially at the solution provider level, to increase their service portfolio.
The use of finely tuned over-dimensional language models on a closed domain does not yet allow for a guaranteed result as can be expected in a support context for example. Wizards based on knowledge graphs can work very well but the problem comes in populating the graphs and maintaining them, which limits their deployment and use.
Supplements
If you would like to learn more about these topics, here is a selection of my other articles:
State Of The Art of Speech Synthesis at the End of May 2021
14 criteria for (well) choosing a chatbots solution
An overview of the global market for chatbot solutions in 2020