Customer Support in the Age of AI: What Can Go Wrong
Customer support is often the first area businesses look to automate. The appeal is obvious: replacing live agents with AI promises cost savings and easier scalability. Many companies jump at this chance to cut staff and reduce expenses, seeing it as a straightforward efficiency gain.
When Automation Fails
Take the example of Klarna, a financial company that replaced hundreds of customer service employees with AI. They expected smoother operations and lower costs, but the reality was different. Customers complained about poor service quality, loss of empathy, and inflexibility. The company had to rehire human agents because full automation failed to meet customer expectations.
Similarly, the Commonwealth Bank of Australia replaced 45 call center employees with a voice bot. Instead of reducing call volume, calls actually increased, overwhelming the system. The bank apologized, restored jobs, and admitted the experiment was a mistake.
Why These Failures Happen
Most support requests are routine and can be automated, but not all. Complex issues often stem from bugs, outdated documentation, complicated systems, or long customer histories. These require nuanced understanding and context that AI cannot easily access.
Customers frequently face challenges when trying to cancel services, switch plans, or understand insurance and warranty terms. These tasks demand human judgment and detailed company knowledge. AI systems lack access to informal knowledge and real-time context that human agents carry.
Outdated IT infrastructure and scattered knowledge bases only make automation harder. Many companies still manage support information in formats like markdown or across disconnected systems, making it difficult for AI to deliver accurate answers consistently.
Rethinking Customer Support
Customer support deals with people who are often frustrated or ready to leave. Solving their problems quickly and efficiently builds loyalty. Yet, many customers dislike contacting support because of past poor experiences.
Support should be viewed as an asset, not a cost to cut. Equip teams with tools that generate auto-responses and documents, and provide access to critical information and context. Instead of focusing on replacing people, focus on improving the quality and speed of support.
Neither humans nor AI can deliver excellent service without the right context. Humans can store informal knowledge and adapt quickly, which AI cannot replicate yet. The goal should be to reduce the volume of support requests through better products and clearer communication, not to heroically handle every case.
Using technology to assist support teams in handling requests efficiently and contextually is the smart path forward. Customer support must be a strategic part of the business, focused on retention and satisfaction rather than just cost savings.
For customer support professionals interested in how AI can enhance their work without replacing them, exploring practical AI training can be valuable. Check out AI courses for customer support roles to learn how to combine human empathy with AI assistance effectively.
Your membership also unlocks:
