There is a pattern playing out across customer service organisations right now. A business deploys an AI tool — a chatbot, a virtual agent, an automated triage system — and the early metrics look promising. Deflection rates rise. Response times drop. The dashboard turns green. Then, quietly, something else happens: customers start to disengage. Satisfaction scores slip. Escalations increase. Trust, which took years to build, erodes in weeks.
This is the tension at the heart of a recent CX Today interview with Dr. Ben Granger, Chief Workplace Psychologist at Qualtrics. His argument is simple and important: the speed of AI adoption in customer experience has far outpaced the pace at which customers are building confidence in it. That gap is not a technology problem. It is a design and deployment problem — and it is entirely avoidable.
What 'Bad AI Rollout' Actually Looks Like
It rarely looks like a catastrophic failure. More often it looks like a series of small friction moments that compound over time. A customer explains their issue to a chatbot, gets a generic response, and has to repeat everything to a human agent. A virtual assistant confidently provides the wrong information about a billing dispute. An automated system closes a ticket that was never resolved. Each incident is minor. Together, they send a clear signal: this company does not have its systems under control.
Dr. Granger's point is that customers are not opposed to AI in service interactions. Research consistently shows that people accept automation when it is fast, accurate, and appropriate to the task. What they reject is AI that overpromises and underdelivers — particularly when the stakes are high, the issue is complex, or they are already frustrated. At that moment, a clumsy handoff or a robotic non-answer does not just fail to solve the problem. It actively damages the relationship.
The Design Principle Most Teams Are Missing
The organisations getting this right are not the ones with the most sophisticated AI. They are the ones that have been most deliberate about where AI operates autonomously, where it supports a human, and where it steps back entirely. That is not a technology decision. It is an operational and customer experience decision.
This means doing the harder work upfront: mapping which interaction types are genuinely suited to full automation, which benefit from AI-assisted human handling, and which require unmediated human judgment. It means training agents not just to use AI tools, but to recognise when those tools are failing a customer in real time. And it means building feedback loops that catch trust erosion before it shows up in churn data.
Most AI deployments skip this work. They start with the technology and retrofit the customer journey around it. The result is exactly what Dr. Granger describes: AI that moves fast, and trust that does not keep up.
Why Hybrid Is the Operational Answer
This is where the hybrid human-plus-AI model stops being a philosophical position and becomes a practical necessity. Not because AI is unready — in the right contexts, it is highly capable — but because customer trust is not uniform. It varies by channel, by issue type, by customer segment, by language, and by the emotional state someone is in when they make contact.
A multilingual customer navigating a complex complaint in their second language needs something different from an English-speaking customer tracking a parcel. A high-value B2B client with a compliance question needs something different from a first-time retail buyer. Blanket automation ignores these distinctions. A well-designed hybrid operation does not.
At Conveneo, this is the operating logic we build from: AI handles volume, speed, and consistency where it genuinely excels — and skilled human agents carry the interactions where trust, nuance, and judgment are what the customer actually needs. The goal is not to minimise human involvement. It is to deploy humans where they create the most value, and AI where it does the same.
The organisations that will win on customer experience over the next three years are not the ones that automate the most. They are the ones that automate the right things — and know exactly where to keep a human in the loop.