St. Louis – October 28, 2026 -- Forty-four percent of U.S. consumers who use AI tools for customer service say the AI has contradicted a company representative or its policies, according to a new survey of 1,000 consumers from support automation vendor Capacity.
The Agentic Consumer Report finds that 55% of respondents have used an AI tool, such as an agent or large language model, to help with a service request. Consumers are equally likely to trust the AI or the company representative when the two give conflicting information.
Gen Z and Millennials drive adoption of personal AI agents
Overall, 35% of respondents have used an AI agent to handle a customer service request at least once. Among Gen Z and Millennials, the figure is 49%.
Household makeup also matters. Among respondents with children aged 25 or younger at home, 46% have used a personal AI agent, against 29% of those without children at home.
AI prompts 34% of consumers to ask more specific questions
Beyond the 34% who say AI makes them ask more specific questions, 23% would ask representatives to explain or cite company policies because of AI. Some 36% would use AI to research return, refund or cancellation policies before contacting a business, so companies may face customers ready to challenge denied requests.
Escalation behavior is changing as well. Twenty percent of consumers said they would escalate issues more quickly when using AI tools. Another 21% would trust their AI agent to connect them with a human representative and then join the conversation.
Peak-season wait times could push 40% of consumers toward AI delegation
About 40% of consumers have used, would use or plan to use a personal AI agent during peak seasons with longer wait times. Thirty percent said they would still not use an agent in those periods. Businesses will therefore need to support both AI-assisted and traditional service interactions, according to the report.
Capacity urges contact centers to build escalation rules and AI-to-AI safeguards
Capacity founder and CEO David Karandish said businesses must move beyond speed and efficiency to ensure answers are accurate and decisions are "defensible."
The report says contact centers will need to equip representatives with accurate company information so they can address discrepancies and stand behind their responses. It also calls for clear escalation processes and safeguards against repetitive exchanges between customer-owned and company-deployed AI agents. Routine requests would stay automated, while human representatives focus on complex issues.
Karandish recommended a continuous learning loop that identifies when consumers are likely to challenge an answer. Insights from those cases would be used to prepare both AI agents and human representatives for more complex, AI-assisted interactions.