AI and Call Centre Workers: What's Actually Happening and What to Do

The honest assessment

i'll be honest. Call centre work is at the sharp end of AI disruption. If you work in a call centre, you already know this. The question isn't whether AI is coming for the role. It's how far and how fast.

The evidence is extensive and it's not encouraging for tier-1 phone and chat support. Klarna's AI assistant now handles two-thirds of customer service chats, equivalent to 700 full-time agents. The company reported that resolution times dropped from 11 minutes to 2 minutes and customer satisfaction remained the same. BT announced plans to shed up to 55,000 jobs by the late 2020s, with customer service among the most affected areas. Vodafone deployed an AI assistant called TOBi that handles millions of customer interactions annually. In the US, major banks and telecoms companies have been steadily reducing call centre headcount while increasing AI chatbot capability.

What AI handles well now: balance enquiries, password resets, order tracking, billing questions, appointment scheduling, FAQ responses, basic troubleshooting, returns processing, and standardised complaint handling. For roughly 60-70% of inbound call centre contacts, the interaction follows a predictable pattern that AI can manage effectively. And the technology is getting better at handling voice calls too. Google's Contact Center AI and Amazon Connect use natural language processing that's increasingly difficult to distinguish from human agents on routine calls.

What AI still struggles with: the call from a recently bereaved person trying to close their spouse's account and breaking down in tears. The complex complaint that spans three departments and six months of botched interactions. The customer who says they want to cancel but really wants someone to listen to their frustration and make it right. The situation where following the process would be technically correct but morally wrong. These moments require genuine empathy, creative problem-solving, and the authority to make exceptions. AI doesn't make exceptions.

But the honest truth is that these complex interactions represent a minority of total call volume. Most call centres exist to handle volume. Volume is what AI does.

Your exposure level: High

High. Among the highest of any role we cover, alongside data entry and basic transcription.

The structural challenge is this: call centres were already built on the principle of standardising and scripting human behaviour. Call scripts. Decision trees. Average handling time targets. Quality frameworks that measure adherence to process. The entire operating model of a traditional call centre is designed to make humans behave as consistently and predictably as possible. AI doesn't need scripting or training to be consistent. It just is. The very architecture of call centre work was always a stepping stone toward automation. AI is simply the technology that makes the final step possible.

The economic incentive is overwhelming. A call centre agent in the UK costs an employer £25,000-£35,000 per year including overheads. An AI agent handling the same volume of basic enquiries costs a fraction of that. When the quality is comparable for routine interactions — and it increasingly is — the business case is straightforward. Companies aren't implementing AI call centre solutions out of curiosity. They're doing it to reduce costs. The language is always "improving efficiency" and "enhancing customer experience." What it means is fewer human agents.

However, "high exposure" doesn't mean "everyone loses their job tomorrow." The transition is happening over years, not months. AI still can't handle everything. Regulation in some sectors requires human agents for certain interactions. And there's growing evidence that the best outcomes come from AI handling the routine work while humans handle the exceptions. The question is whether you position yourself as the human who handles the exceptions, or whether you're the one doing the routine work that AI is about to take over.

The 90-day action plan

1. This week: learn to use every AI-assisted feature in your current systems. Most modern call centre platforms — Genesys, Five9, NICE, Salesforce Service Cloud — have AI features. Suggested responses. Auto-summaries. Sentiment detection. Knowledge base recommendations. Use them all. Understanding these tools makes you more efficient now and more valuable in an AI-augmented team later.

2. Week two: deliberately seek out the difficult calls. When a complex escalation comes through, don't groan. Volunteer for it. Every difficult interaction you handle well is evidence of your value beyond what AI can do. Start keeping a log of your complex case resolutions. Note what made each one difficult and how you resolved it.

3. By day 30: develop specialist product or process knowledge. Become the person on the team who really understands the product, the billing system, or the complaints procedure. Deep knowledge allows you to handle edge cases that AI can't. It also positions you for quality assurance, training, or specialist escalation roles.

4. By day 45: learn quality assurance and coaching skills. Start thinking about what good looks like in customer interactions. If you can evaluate and improve the performance of both human agents and AI systems, you have a role that sits above the automation. Offer to help with new starter training or call quality reviews.

5. By day 60: explore the adjacent roles within your organisation. Customer success. Account management. Complaints handling at a senior level. Operations. Workforce management. Training and quality. Talk to people in these departments. Understand what skills they need. Start building those skills alongside your current role.

6. By day 75: build data literacy. Call centres generate enormous amounts of data — call volumes, resolution rates, satisfaction scores, first contact resolution, escalation patterns. Learning to analyse and interpret this data is a skill that transforms you from a call handler to someone who understands the operation. Basic Excel analysis, understanding of key metrics, and the ability to spot trends are all valuable.

7. By day 90: propose your next role. Go to your manager with a plan. "AI is going to handle more of the routine calls, and that's the right move for the business. I'd like to specialise in [complex escalations / quality assurance / training / operations support / customer retention]. Here's what I've been doing to prepare." You're not asking permission to keep your current job. You're proposing the evolution that makes you indispensable.

AI tools you should be using this week

• ChatGPT for Work — Use it to prepare for difficult interactions. Describe a complex customer scenario and ask for suggested approaches. Practice how you'd handle an angry customer about a specific issue. Also useful for drafting follow-up emails, understanding product details you're not sure about, and preparing for performance reviews.

• Microsoft Copilot for Work — If your organisation uses Microsoft 365, Copilot can help you manage email follow-ups, summarise long case histories quickly, and create reports on your case resolutions. Useful for building the documentation of your work that supports your case for progression.

• Grammarly AI — Essential for any written customer communication. Checks tone as well as grammar. In written customer service, the difference between a message that resolves and a message that escalates often comes down to tone. Grammarly catches the phrasing that might accidentally come across as dismissive or condescending.

• Claude for Work — Good for working through complex customer problems methodically. "A customer has been billed incorrectly for three months, has called four times, and is threatening to go to the ombudsman. Here's the history. What's the best resolution approach?" Claude will walk through the situation step by step, which can help you prepare for challenging interactions.

What to say in meetings

When management announces AI chatbot expansion: "I think that makes sense for the routine enquiries. What I'd like to understand is the plan for the interactions AI can't handle well. The complex cases, the emotional situations, the ones that could become complaints. I'd like to be involved in designing the escalation process and handling those cases."

If colleagues are anxious about job losses: "Some of the routine call work is going to AI. That's happening. But look at it this way — nobody got into this job because they love answering the same question 200 times a day. The opportunity is to move into the work that actually requires a human brain. Escalations, retention, quality, training. Those roles aren't going anywhere."

In performance reviews: "My complex case resolution rate is [X%] with a [Y] satisfaction score. These are the cases AI can't handle. I've also been developing my skills in [quality assurance / training / data analysis]. I'd like to discuss how my role can evolve as the AI capability expands."

If the worst happens

If you're made redundant from a call centre role, take a breath. Your skills are real. You can communicate under pressure. You can de-escalate conflict. You can multitask across multiple systems while maintaining a conversation. You can follow complex processes while adapting to unpredictable people. These skills transfer to sales, recruitment, office management, reception, account management, insurance claims handling, and any client-facing role.

Adjacent roles to consider: customer success manager (particularly in SaaS companies, where the role is essentially strategic customer service), sales development representative, recruitment consultant, insurance claims handler, complaints officer in financial services or healthcare, or operations coordinator. Many former call centre workers also move into team leadership, training, and workforce management roles within the contact centre industry itself — the industry is shrinking at the agent level but still needs managers, trainers, and quality professionals.

Here's something i believe strongly. Call centre work is undervalued and often looked down on, which is rubbish. Dealing with angry, confused, or upset people all day while simultaneously navigating complex systems and meeting performance targets is genuinely skilled work. When you're describing your experience, don't minimise it. "I resolved an average of [X] customer issues per day with a [Y%] first contact resolution rate and a [Z%] satisfaction score" is a powerful statement. Own the skills. They transfer more broadly than you think.