AI tools are now part of the daily life of many customer relations advisors, but their effects are not the same depending on the profile of the employees. This is what the first edition of the Observatory on the Impact of AI in Customer Relations, published by Smart Tribune and the blog Sens du Client, shows. Thomas Gouritin, AI expert at Smart Tribune and coordinator of the Observatory, comments on its main findings.
Three top uses, perceived benefits more modest
Three uses come out on top, regardless of the respondents’ profile: response assistance, with suggestions and rephrasing; automatic summarization of exchanges after a call or chat; and the intelligent knowledge base. Each of these is used by 55.4% to 65.5% of participants, depending on the time they spend with their clients.
The way of working and team productivity are viewed rather favorably, with scores around 5.5 to 5.8 out of 10. The concrete benefits, rated from 0 to 5, are however more modest. The sense of personal efficacy comes first with 2.56, while improving customer satisfaction only gets 2.25. The ability to perform tasks previously out of reach receives the lowest score. According to the report, advisors primarily need tools that facilitate handling customer requests.
For Thomas Gouritin, a useful AI almost makes itself forgotten. It must provide the right answer at the right time, without forcing the advisor to leave the conversation to search for information elsewhere. It must also be reliable, because a tool that constantly needs rechecking wastes time and ends up not being used. This reliability depends, in his view, less on the technology than on the quality of the knowledge that feeds it.
Frontline advisors less convinced
More than half of respondents (53.8%) spend over 75% of their time in direct contact with clients. On each of the five concrete benefits measured, these frontline advisors assign lower scores than other participants, with an average of 2.26 out of 5 versus 2.44. In this group, zero is even the most frequent score, and more than one in ten advisors reports zero operational impact.
Their use of AI is also more varied. They are more likely to use it for most of their tasks (16.2% vs. 8.2%), but also to use it only rarely or never (30% vs. 22%). Nearly a third of respondents (31%) resort to personal AI tools, outside the framework set by their company, a practice called ‘Shadow AI’. It is less prevalent on the frontline, where it concerns 26.4% of advisors, compared to 36.5% for the rest of the panel. The report sees this as the effect of live interactions, where the risk of error or waiting during the conversation discourages using an external tool.
According to Thomas Gouritin, this gap reveals a disconnect between tool design and on-the-ground reality. An AI conceived far from the teams often adds a screen, a response to check or rephrase, and when the information feeding it is incomplete or outdated, it’s the advisor who faces the consequences with the client. He specifies that the gap does not reflect resistance to change, which the Observatory confirms.
Human skills at the heart of customer relations
When respondents express themselves freely, one in five answers (20.3%, out of 281 analyzed responses) values AI for the time it frees up for the benefit of clients. Others express reservations, with 17.8% of responses indicating rejection or strong concern, particularly regarding the risk of replacement. The security and confidentiality of customer data top the list of risks, cited by 74.5% of frontline advisors and 74% of other respondents.
Katia Grenu, Head of Customer Relations at Alltricks, testifies in the Observatory: ‘Future employability will largely depend on what AI cannot do: negotiation, reassurance, understanding context, emotional personalization of contact.’ For companies, strengthening these skills involves training, among other things. Thomas Gouritin believes that the advisor will occupy a more central role as AI handles simple requests, and that they can also contribute to the quality of the tool by detecting false answers or missing information.
The methodology is based on a survey conducted by Smart Tribune, which markets AI solutions for customer relations. The questionnaire was distributed remotely between July and September 2026, partly among Smart Tribune’s clients. It gathered 435 professionals, including 73.8% advisors, agents, or technicians in contact with customers, and 18.5% customer service managers.
