Innso Day brought together on September 19 in the heart of the “Silicon” Sentier Parisien in the Mail district, more than 70 professionals and experts in Customer Relations and digital innovation. During a round table dedicated to automation and AI, Guillaume Laskowski and Karim Ben Djemiaa, Associates Partners at EY France as well as Marco Noel, Watson IBM Offering Manager, alongside Innso CEO Nicolas Raffin , demystified the dream (or nightmare) sold with automation and AI technologies by exposing through concrete cases the real gains, as well as the limits.

Man and machine: a tug-of-war or of flesh?

A McKinsey study published in 2017 predicted a 14% replacement rate of jobs worldwide due to AI. Faced with this disaster scenario, a 2018 PwC report tempered these forecasts, indicating that by 2030 AI will have generated as many jobs as it will have eliminated.

Representing a market estimated at 149 Million euros by 2022, with 70% of companies that have already started deploying these technologies, one thing is clear: AI and RPA (robotic process automation) must find their place, but they will not replace the human, nor the expert.

Their association offers advantages that no one can ignore: savings – 25-35% which can go up to 70%, efficiency (acceleration of processing times, reduction in the number of errors) and above all enrichment of the customer relationship. Because, beyond cost reduction, the fact that the most tedious and time-consuming tasks are automated does not spell the end of the profession of client advisor, on the contrary. It will be able to refocus on its real added value: the personalization of the customer relationship and targeted expertise. The customer experience will be improved and the turnover rate will be compressed.

The Innso solution follows this logic. “We are in a process of increasing the capacities of advisers. The area for improvement is to help advisors better understand clients’ needs. ”explains Nicolas Raffin.

Demystify by demystifying acronyms

RPA is a software solution that has two major characteristics, explains Guillaume Laskowski: It fits into the existing application environment, and reproduces what humans do but in a more rapid and systematic way – the robot is not intended to create things on its own.

Process, consolidate and export information from a series of Excel files received by email, for example, now the RPA can do it for the employee. Or, in the case mentioned of a major European telecommunications player, arrange to automate more than 205 clicks, 60 windows to open in addition to a Notepad which manages all the copied and pasted information, this which divided by 6 the processing time of requests.

“RPA is there not to make humans a robot but to relieve them of all mechanical and repetitive tasks,” emphasizes Karim Ben Djemiaa. “There is a quantity of processes on which we can gain in productivity, which contributes to their overall profitability. This type of technology does not involve a major overhaul of existing systems, an adaptation is certainly necessary but it remains extremely light as a program with very short ROI: less than a year, and a rapid implementation – between 6 and 16 weeks, depending on the level of complexity and the number of robots. ”

AI is a collection of intelligent services , including among others, artificial vision, algorithms, machine learning, natural language recognition … which allow the intentions of the user to be identified. user, extract key insights and then set up a decision-making process based on these insights. »Explains Marco Noël.

In an environment where tasks can be automated because the business rule is known, RPA is fine. On the other hand, when the information is “moving” and it is a question of detecting the intention of the customer within an unstructured document (such as the email of a customer who parasitizes its claim with “semantic noise”, in other words, only formulates it after ten lines) and put it in correspondence with a product among a multitude, then artificial intelligence can complete this automation.

Deployed for Le Crédit Mutuel, IBM Watson has enabled advisers to process the 35,000 emails received by the bank twice as quickly and respond more quickly to 60% of the questions asked on more than 200 products, with the sending of automatic responses for simple intentions and a response aid for more complex intentions.

Customer service is not purely and simply improved: thus optimized, it can offer new services to customers. Beyond operational efficiency , there is the experiential dimension, adds Nicolas Raffin: “A whole research exists around AI to meet the expectations of a new generation of customers at the level UX – in particular with the new conversational channels which promote a much richer relationship with the customer ”.

But not at any cost. Whatever the ambitions, it is above all important to start from the field analysis of a specific use case, and to involve users in order to properly define expectations : “You have to be sure to meet 100% the requirements of your users; because if the expectations are enormous, the risk of disappointment is also immense ”, warns Marco Noël.

RPA or IA how to choose?

“Go out into the field, on your sets, and observe the number of clicks, the frustration felt by your advisers and all the dispersion of your resources” recommends Nicolas Raffin. Then, once you have identified your problem and precisely defined your use cases and your objectives, it is a question of convincing senior and middle managers, advises Guillaume Laskowski, by communicating clearly at all levels. of the company, and by concretely demonstrating via “botathons”, pocs, pilots … all the exploitable potential of this technological wave.

The objective being to support the customer as well as possible by optimizing processes and bringing the desired added value to the customer relationship, the solution involves closely examining the customer experience and that of the employee </ em > to consider the best possible combination of RPA, AI and human expertise. In some cases, it must be said, depending on the level of maturity of the company, it will suffice to upgrade the IS.

To relive this highlight of the day, we offer you a replay of the innovation round table.

moderated by Nicolas Raffin , CEO of innso, in the company of:

Marco Noel , responsible for the Watson offer at IBM

Guillaume Laskwoski and Karim Ben Djemiaa , Associate Partners at EY Consulting