Gartner Survey Finds 87% of Customers Say Companies Using GenAI for Customer Service Must Provide Access to a Human Agent

What happened

Gartner has published new survey findings showing that 87% of customers say organisations deploying generative AI in customer service must still offer a clear route to a human agent. The research signals that as companies expand GenAI-powered chatbots and virtual assistants, customers are not rejecting automation outright — but they are drawing a hard line around the ability to escalate to a person when needed.

The finding lands as more service organisations move GenAI from pilot to production across contact centres, self-service portals and messaging channels. Gartner's data suggests that the absence of a visible human fallback, rather than the presence of AI itself, is what risks damaging customer trust.

Why it matters

For CX leaders, this reframes the automation debate away from "AI versus human" and towards "AI plus guaranteed human access." Behavioural economics has long shown that perceived loss of control is a powerful driver of dissatisfaction; when customers sense they cannot reach a person, anxiety and distrust spike even if the AI itself performs adequately. Escalation paths function as a psychological safety net, not just an operational backstop.

Service design implications follow directly: the fallback needs to be easy to find, not buried behind menus or evasive prompts. Organisations that treat human access as a genuine design requirement — visible, low-friction, and honestly communicated — are more likely to sustain trust as they scale GenAI. Those that treat it as an afterthought risk a backlash that undermines the efficiency gains automation was meant to deliver.

By the numbers

• 87% of customers say companies using generative AI for customer service must provide access to a human agent, according to Gartner's survey.

The Renascence take

The headline number is less about AI scepticism and more about control and dignity in service encounters. Customers are pragmatic: they will accept automation for speed and convenience, but only if they retain agency over the interaction.

Most operators will read this as a mandate to "add a chat-to-human button" and call it done — that misses the point entirely. The real design question is how visible, how fast and how honest that escalation path is at the moment a customer feels stuck, not whether it exists in a footnote or FAQ. Brands that quietly make human access harder to find than the AI itself are optimising for cost at the expense of trust, and customers now clearly notice the difference. The winning move is to treat escalation transparency as a core metric alongside containment rate — not a concession to be minimised.