Will AI Replace Call Centre Workers? The Numbers Are Already In

i'm going to be more direct in this one than i usually am, because i think you deserve honesty over comfort.

Klarna replaced the equivalent of 700 full-time customer service agents with an AI chatbot. Not in theory. Not in a pilot programme. In production, across 23 markets. They reported it handled two-thirds of their customer service conversations in its first month. Resolution time dropped from 11 minutes to 2. Customer satisfaction stayed the same.

BT announced plans to cut 55,000 jobs by the end of the decade, with a significant portion in customer service roles being replaced by AI.

These aren't hypothetical scenarios. These are things that have already happened. And if you work in a call centre, you don't need me to tell you that the atmosphere has changed. You can feel it.

i was made redundant from a data science position. The dread you might be feeling right now? i know it. The knot in your stomach when you hear about another company announcing AI customer service? i know that too. So let me give you the honest picture, including the parts that aren't great, and then we'll talk about what you can actually do.

The short answer

Call centre work is among the most exposed professions to AI automation. This is not a nuanced "well, it depends" situation for a large portion of the industry. Routine, scripted, high-volume customer interactions — the kind that make up the majority of call centre work — are being automated now, at scale, by companies that are publishing the results and showing it works. Some human roles will remain for complex cases, escalations, and situations requiring genuine empathy. But the overall headcount in call centres is going to fall significantly over the next few years. i wish i could tell you something different.

What AI can already do in call centres

The capabilities are further along than most people outside the industry realise.

Voice AI systems can now handle phone calls — actual phone conversations — with a quality that most callers can't distinguish from a human agent. The "press 1 for billing" menu is being replaced by conversational AI that asks what you need and handles it. Not perfectly, not for everything, but for a growing range of interactions.

Chat-based customer service is where the automation is most advanced. Chatbots powered by large language models can now handle nuanced conversations, access customer records, process transactions, issue refunds, change account details, and resolve complaints — all without a human touching the interaction.

Email and ticket handling. AI systems can read incoming emails, understand the request, pull up the relevant account information, and draft or send responses. For straightforward queries, this is fully automated. For complex ones, the AI drafts a response for a human to review and send.

Sentiment analysis in real time. AI monitors conversations and flags when a customer is becoming agitated, frustrated, or likely to churn. This is used both to escalate to human agents and to adjust the AI's tone and approach.

Quality assurance and compliance monitoring. Instead of a team leader listening to a random sample of calls, AI monitors every single interaction for compliance, quality, and policy adherence. This both improves quality and reduces the management overhead.

Multilingual support. AI can now handle customer service in dozens of languages without the company needing native-speaking agents for each market. This is what allowed Klarna to deploy across 23 markets simultaneously.

What AI still can't do

There is still a human layer, and it matters. Even if it's smaller than before.

Complex complaint resolution that involves genuine empathy. The customer who's been passed around for weeks, is genuinely upset, and needs someone to take ownership and fix the problem. AI can process the complaint. It can't make the customer feel heard in the way a skilled human can.

Situations with high emotional stakes. Bereavement cases at insurance companies. Customers in financial distress. Vulnerable people who need patient, adaptable communication. These interactions require the kind of emotional intelligence that AI simulates but doesn't possess.

Edge cases and novel situations. The request that doesn't fit any existing process. The complaint that reveals a systemic issue nobody knew about. The customer whose situation is so unusual that there's no script for it. These still need human judgement and creativity.

Negotiation and retention. When a high-value customer wants to leave and the company wants to keep them, the conversation requires reading between the lines, making judgement calls about what to offer, and building a personal connection. The best retention agents are essentially salespeople, and that skill set is harder to automate.

Regulatory and legal complexity. In financial services, healthcare, and other regulated industries, some customer interactions have legal implications that require human accountability and professional judgement. AI can't sign off on regulated advice.

The real risk

i'll give it to you straight.

The call centre industry employs millions of people globally. A significant percentage of those roles — estimates vary, but 40-60% is the range i see most often in the restructuring conversations i'm part of — are at high risk of automation within the next three to five years.

The pattern i see is this: companies start with AI handling the simple queries. Then they expand to medium-complexity interactions. Then they use AI to assist human agents on complex calls. Then they reduce the human agent headcount because each remaining agent, augmented by AI, can handle more. The endpoint isn't zero humans, but it's far fewer than today.

Offshore call centres are particularly vulnerable. If the cost advantage of offshore labour disappears because AI is cheaper than any human workforce, the economic model of the offshore call centre industry collapses. This affects millions of workers in India, the Philippines, and other BPO hubs.

The speed of this transition is what makes it particularly brutal. This isn't a gradual shift over a decade. Companies are moving fast because the economic incentive is massive. Call centres are expensive. AI is relatively cheap. The maths is simple and the results are provable.

The jobs that remain will be different. Higher-skilled, higher-stakes, requiring more expertise and emotional intelligence. There will be fewer of them, and they'll likely pay better than current call centre roles. But the transition from "lots of jobs at current skill levels" to "fewer jobs at higher skill levels" is going to be painful for a lot of people.

What to do about it

i'm not going to insult you with "just upskill" as if it's that simple. But here's what i'd actually do if i were in a call centre right now.

1. Start moving toward specialisation now, not later. Within your current role, push toward the complex cases. Volunteer for the difficult customers, the escalations, the complaints that nobody else wants to handle. These are the interactions that will still need humans. Build a track record in them.

2. Target the industries where human contact is regulated or essential. Financial services, healthcare, insurance, legal services — these sectors still need human agents for compliance reasons and will for longer than others. If you're in a general customer service call centre, consider moving to a regulated industry where the human requirement is built into the rules.

3. Develop adjacent skills that call centres build naturally. You already know how to communicate clearly, handle difficult people, work under pressure, manage time, and navigate complex systems. These are transferable skills. Sales, account management, customer success, complaints resolution, training — all of these are careers that build on what you already do.

4. Learn the AI tools, don't just resent them. The people who manage AI customer service systems, train them, monitor their quality, handle their escalations, and improve their performance are a new and growing workforce. Understanding how AI customer service works puts you ahead of people who only know the old way.

5. Have an honest conversation with yourself about timeline. If your specific role is primarily handling routine queries by phone or chat, the window is shorter than you'd like. Don't wait for the restructuring announcement. Start your transition now, while you're still employed and can make choices from a position of stability rather than desperation.

The bottom line

i don't enjoy writing pieces like this one. Most of the "will AI replace X" articles i write have genuinely nuanced answers. This one has a less comfortable answer: for a large portion of call centre work, yes, AI is replacing it, and it's happening now.

But here's what i keep coming back to: the skills that make a good call centre worker — communication, patience, problem-solving under pressure, the ability to deal with angry humans all day without losing your mind — those are genuinely valuable skills. They transfer to dozens of other roles. The job might be changing, but the capabilities you've built aren't worthless.

The worst thing you can do is freeze. The second worst thing is pretend it's not happening. The best thing is to acknowledge the reality, identify where your skills have value in a changing market, and move before you're forced to. It's not fair that you have to. But fair and real aren't the same thing. i learned that the hard way.