Years of bad chatbots have left customers expecting AI customer service to fail, Gradient Labs data suggests

Consumer patience with AI customer service has eroded to the point where 'bypass chatbot' is now one of the fastest-growing search queries online. Gradient Labs, which builds autonomous AI agents for fintechs including Wise, Current, and Zego, commissioned a survey that captures the scale of the problem — and uses it to argue that the issue isn't AI itself, but the wrong kind of AI being applied to high-stakes financial queries.

Dimitri Masin, CEO and co-founder of Gradient Labs, places the blame on horizontal, general-purpose chatbots that were designed to answer frequently asked questions rather than complete regulated, multi-step processes inside a bank's systems. "When the first chatbots came out, they weren't built to fix customers' issues, but to answer common questions and redirect them elsewhere. That's why, after the third repeat, customers scream 'let me speak to a human' — the person who could actually understand and help," he said.

The company's argument is that vertical AI agents — each built for a specific regulated workflow, such as the borrower lifecycle, dispute management, or identity verification — can outperform both general chatbots and human agents on customer satisfaction and quality assurance metrics. Masin said that at Gradient Labs' fintech customers, these agents are already running continuously, sharing context and memory and passing work between systems.

"What makes that possible is having a suite of agents, each built for one regulated job, automating the complex, long-running workflows a general-purpose agent can't take on," he said. "This isn't a future of agentic AI — it's already happening."

The consumer research adds data to a debate that has been running for several years in financial services: whether AI in customer service is a cost-saving measure that degrades experience, or a genuine improvement when implemented with sufficient domain specificity. Gradient Labs has a clear commercial interest in making the case for the latter.

To stay across the latest in cloud, AI and enterprise tech analysis from Compare the Cloud, subscribe to our weekly newsletter at https://www.comparethecloud.net/newsletter

More News