AI transforms customer relations without replacing humans. The hybrid model combines speed, availability and empathy to provide a more efficient and human experience.
By 2027, theartificial intelligence could handle half of the requests sent to customer services. However, consumers remain overwhelmingly attached to human advisors. This paradox does not condemn AI: it confirms that the most efficient model will probably be hybrid.
Artificial intelligence is no longer an experiment reserved for a few large companies. It is gradually being established in customer services, contact centers and support pathways.
According to the State of Service 2025 report from Salesforceprofessionals surveyed estimate that AI already handles around 30% of customer service cases. This proportion could reach 50% by 2027. The study is based on responses from 6,500 customer service professionals around the world.
This progression responds to a very real expectation.
According to the CX Trends 2026 report from Zendesk74% of consumers now expect customer service to be available twenty-four hours a day and 88% want faster responses than a year ago.
Companies must therefore respond faster, on more channels and during ever-expanding time slots.
In this context, artificial intelligence seems to provide an obvious answer: it is permanently available, can absorb a large number of requests simultaneously and does not experience peaks of activity on Monday morning.
But customer expectations are about more than speed.
The more technology advances, the more customers demand human touch
The 2025 Customer Services Observatory, produced by Ipsos among 5,000 European consumers, reveals a particularly interesting paradox.
In France, 90% of people surveyed say they prefer to wait to speak to a human advisor rather than speaking immediately with a virtual advisor.
Furthermore, 31% of French people already use chatbots, but only half of them say they are satisfied with this channel. Finally, 72% would feel like they had been deceived if a company did not inform them that they were trading with artificial intelligence.
These figures could be interpreted as a rejection of AI.
That would be too hasty a conclusion.
Consumers do not necessarily reject technology. They especially refuse devices that waste their time, force them to repeat their request or prevent them from accessing a person when the situation becomes complex.
The problem is therefore not artificial intelligence itself.
The problem arises when it is used as a barrier rather than a service.
The mistake is wanting to automate all conversations
Not all customer service requests have the same value or level of complexity.
A customer who wants to know the opening hours of an agency, check the status of an order or modify an appointment is mainly looking for a quick response.
In this type of situation, AI can fulfill its mission perfectly.
It can identify the customer, find the necessary information, qualify the request, propose an action or direct the call to the right department.
It thus prevents consumers from waiting unnecessarily and allows human teams to concentrate on more complex subjects.
But a conversation with a customer is not always limited to a succession of questions and answers.
Behind a seemingly simple request may hide concern, dissatisfaction, misunderstanding or a business opportunity.
A client who asks about the status of their file is not just looking for a date. He may be worried because this answer conditions an important project.
A prospect asking for a price doesn’t always just want to get a number. He may need to be reassured, advised or supported in his decision.
An unhappy customer doesn’t just expect a technically correct answer. He also wants to feel like his situation has been understood.
It is precisely in these moments that humans retain a decisive advantage.
Artificial intelligence processes an intention, humans understand a context
An experienced advisor knows how to detect hesitation, reformulate a request, adapt their vocabulary and hear what is not explicitly formulated.
He can also go off-script when the situation calls for it.
Artificial intelligence is particularly effective when it operates within a defined framework. Humans remain essential when the conversation requires discernment, empathy, negotiation or taking responsibility.
A study published in February 2026 by McKinsey, conducted among 440 executives and customer relations managers, shows that nearly 70% of respondents consider that empathy and trust will always require human intervention.
This conviction does not mean that customer relations professions will remain unchanged.
On the contrary, advisors will probably have to deal with fewer simple requests, but more complex, sensitive or strategic situations.
Their role will gradually evolve from transmitting information to resolving, advising and creating trust.
The best model does not pit humans against AI agents
The debate on replacing humans with AI is probably ill-advised.
The real question is not which one should win. It consists of determining how to intelligently distribute roles.
In a hybrid model, AI can support the first level of contact:
identify the reason for the request;
authenticate or recognize the client;
collect initial information;
answer simple questions;
propose a meeting;
carry out certain operations;
guide the conversation.
When the situation becomes more complex, emotional or strategic, the client must be able to be transferred to a counselor.
This transfer should not be considered a failure of automation.
On the contrary, a good AI agent is also one who recognizes the limits of its intervention and knows how to hand over at the right time.
The advisor must then retrieve the history of the conversation, the information already communicated and the actions previously carried out.
The customer should not have to start their journey from the beginning again.
It is this continuity that transforms a juxtaposition of technologies into a true customer experience.
AI must also work for advisors
The hybrid model is not just about placing an AI agent before a human advisor.
Artificial intelligence can also support the employee during and after the exchange.
It can search for information in a documentary database, suggest a response, summarize a conversation, detect a feeling, suggest the next action or automatically complete a report.
In this configuration, the AI does not replace the advisor: it becomes his co-pilot.
The employee spends less time on research and administrative tasks. He can focus more on listening, explaining and solving.
This approach can also improve working conditions.
In customer relations centers, part of the fatigue comes from the repetition of requests, the multiplication of tools and the need to enter a lot of information after each interaction.
Automating these tasks can restore time and value to human work.
But to obtain this result, the teams must be associated with the project. Artificial intelligence imposed solely as a cost-cutting tool risks causing distrust.
AI presented as a way to remove repetitive tasks and improve the quality of work will be more easily adopted.
Performance does not only depend on the automation rate
Companies often measure their artificial intelligence projects through a few indicators: number of automated requests, average processing time, transfer rate or cost per contact.
This data is useful, but it is not enough.
A business can automate more while degrading its customer experience.
It can shorten the average conversation duration while increasing the number of callbacks.
It can reduce the cost of each interaction but increase unresolved requests.
The automation rate should therefore not become the main objective.
The essential indicators remain first contact resolution, satisfaction, repeat rate, the effort required of the customer and the ability to detect situations requiring human intervention.
McKinsey’s work also shows that the most advanced companies do not simply install more technologies.
Among organizations considered leaders in customer relations, 40% say they have significantly improved their customer experience scores over the last twelve months, compared to only 12% of the least advanced companies.
The difference is not just about access to AI. It is mainly due to the ability of these companies to review their processes, their organization, their skills and their governance.
Before automating, you need to understand the conversations
A company should not start its project by choosing a chatbot, a voice agent or a technology platform.
She should first analyze her customers’ demands.
What are the main reasons for contact?
Which requests are actually repetitive?
Which exchanges require expertise or negotiation skills?
When do clients ask to speak to an advisor?
Which paths generate misunderstandings, abandonments or recalls?
This analysis makes it possible to distinguish three categories:
requests that can be fully automated;
requests that can be prepared or assisted by AI;
requests must remain primarily human.
It also avoids a common mistake: automating a process that already works poorly.
Artificial intelligence does not spontaneously correct poor organization. It can even accelerate malfunctions.
Before automating, we must therefore simplify, clarify and structure.
The future will be hybrid, but above all well orchestrated
The adoption of AI by French companies is accelerating.
According to the OhMyCX Observatory! 2025 from Ipsos, the integration of artificial intelligence is cited as a priority by 26% of companies surveyed, compared to only 7% in 2023. At the same time, improving the customer experience remains their first strategic priority, mentioned by 53% of organizations.
These two objectives should not be opposed.
Artificial intelligence can improve the customer experience when it instantly responds to a simple request, efficiently prepares for a conversation, or helps an advisor better resolve a situation.
It degrades it when it seeks to prevent human contact, hides its identity or locks the customer into a dead-end path.
The most successful organizations will therefore be those that know how to automate simplicity, assist complexity and preserve humanity when listening, empathy and discernment make the difference.
The future of customer relations will be neither totally human nor totally artificial.
It will be hybrid.
And its success will depend less on the power of the technology than on the intelligence with which companies organize collaboration between their AI agents and their employees.
Fabien Ledoux, president and founder of ContactMedia, a customer relations center specializing in the outsourcing of customer exchanges and hybrid solutions combining human advisors and AI agents.
