Customers increasingly ask ChatGPT and similar tools before using a company's own chatbot. A Gartner survey found that customers now use external GenAI applications such as ChatGPT, Gemini or Claude around three times as often as company-provided chatbots when they need support.
For companies, this is more than a new channel. It changes where customer service begins and creates new requirements for customer service management.
The Gartner survey of 3,566 B2B and B2C customers was conducted in February and March 2026.
Notable shifts:
- Customers use external GenAI systems around three times as often as company-provided service chatbots.
- The use of external GenAI tools for service has almost doubled within a year, while company chatbot usage has remained statistically largely unchanged since 2022.
- 58% of GenAI users had already asked AI to complete tasks for them. In B2B, this figure was 74%.
- At the same time, customers still expect the option of being handed over to a human employee.
The first contact increasingly happens outside the website
For a long time, the company website was the main starting point for service requests. Customers looked for the FAQ, opened the chatbot or wrote an email.
Today, that first step often looks different:
- “How do I make a complaint to company X?”
- “What rights do I have?”
- “How long does processing take?”
- “Where can I find the right form?”
- “What refund am I entitled to?”
The answers increasingly come from ChatGPT or other AI assistants, rather than the company itself.
For companies, this means having less control over the entry point into service processes. They need to ensure that their information is understandable, up to date and reliably interpretable by AI systems.
Competition is shifting
Many companies continue to invest in chatbots. However, Gartner's findings show that the nature of competition is changing.
The key question will no longer simply be:
“Who has the best chatbot?”
Instead:
“Who provides the most reliable and efficient service, regardless of where the request originated?”
A chatbot can provide information. It does not, on its own, resolve complex service processes.
AI changes the entry point, not the process
When ChatGPT recommends that a customer submit a complaint or an application, the real work of customer service begins.
Processes then need to work:
- AI-assisted case recognition and creation
- Intelligent routing
- Clear responsibilities
- SLA and deadline management
- Escalation workflows
- Audit-ready documentation
- Transparent communication
- Analysis and reporting
For most companies, it should initially make little difference whether a request originated via ChatGPT, email, telephone or a web form.
Customer service management becomes a key success factor
Every additional communication channel increases complexity.
Customers expect companies to respond quickly across channels. They are not interested in which systems are used internally. They expect their request to be handled quickly, clearly and professionally.
This makes customer service management increasingly important.
A central service platform helps ensure that all cases are handled according to the same rules, regardless of the incoming channel. Information remains complete, deadlines can be monitored and each processing step is documented transparently.
Especially in regulated industries, reliable processes are a prerequisite for compliance and service quality, rather than simply a convenient feature.
The role of AI is changing
The Gartner survey highlights another important trend: customers increasingly use GenAI to get tasks done, rather than just to find information.
A supporting technology is becoming a central service channel.
AI will increasingly accompany the first steps in customer service. However, quality is determined by what follows: the processes, workflows and handling behind the scenes.
External AI systems may increasingly take over the first customer contact. Companies have limited control over that development, but they can control the quality of their internal service processes.