By Luke Sartain, CEO and Founder of Narwhal Labs
AI and jobs is a topic I’m living right now, not just commenting on. We’re building the infrastructure that’s literally automating customer-facing roles at scale – voice agents handling inbound calls, SMS, email – so I’ve got a pretty clear view of what’s actually changing versus what’s hype.
To put it honestly, any job that requires the use of a phone or computer is at risk. A report released by the Greater London Authority the other day claimed one fifth of London jobs were at risk of AI. This is a significant underestimate. From our observations, within the decade, we could see four in every five white-collar jobs performed by AI.
But this isn’t about robots taking over. Instead, it’s about specific task layers disappearing first – and the UK’s concentration of financial services, professional services and call centre-heavy industries makes it more exposed than most.
Anyone working in business, and any young person applying for a new job, can see that entry-level and junior roles are the real casualty, not senior ones. AI is already performing tasks effectively that used to be “learning the ropes” for new joiners – anything from handling inbound calls and acting as the first point of contact to research and data collection.
As a result, we’re seeing hiring patterns already shifting. Companies are hiring fewer people but expecting more from each of them – employees who can use AI to enhance their roles. But there is a distinction between replacement and automation. Most jobs aren’t disappearing directly, they’re being restructured around whoever can work with AI. The key point to remember is that outperforming someone is not the same as replacing them, and that society must deliberately choose what must never be replaced by machines.
A good way to outline what’s happening is to use examples from what we’re deploying ourselves in industries like financial services and customer operations. The use of AI is being driven by rising customer expectations and costs. In banking, for example, the majority of customers expect rapid responses to their queries, but the average response time is 4.2 hours. Additionally, poor service contributes to hundreds of millions of pounds in UK bank complaints costs annually.
AI is uniquely equipped to match these challenges. Now, AI agents can instantly handle account queries, book appointments and escalate fraud alerts over a phone call, SMS or email at any time. If a customer can receive answers to their queries immediately or until a human has capacity to deal with a query in need of escalation, then you are far more likely to prevent customer churn.
As AI agents continuously learn from every interaction, their ability to respond to requests and incidents like fraud will only get better. They are starting to do this not only within a single call, but across weeks of interactions and whole campaigns, where every conversation builds on the last.
Given research by the Bank of England and the FCA showed that three quarters of UK banks had already adopted AI in some form in 2024, and a further 10% planned to adopt it in the next three years, you can see how quickly the sector has implemented the technology. It shows that front office jobs are being automated, but also the value of maintaining human teams for the escalation of cases.
These growing capabilities will stop the need to hire new team members into branch and CX teams. Over time, there will be smaller teams who are only required to handle the most sensitive and complex cases. And these efficiency gains can be delivered across sectors, reducing entry level and front office workers.
To stay ahead, companies and individuals need to learn how AI can allow them to work better. They need to rethink what entry-level roles look like and how new starters can learn the skills they need. Any company deploying AI at scale should be mandated to invest in reskilling and redeploying any affected workers.
There should be clear rules on where AI can and cannot replace human roles, particularly in care, education and public safety. The truth is all middle management and below are at risk, from receptionists and accountants to web developers and lawyers. There is also future humanoid displacement too for roles such as shelf stackers and checkout assistants. That’s when we start looking at four in five jobs being automated.
Individuals can and should learn how to use AI to their advantage. But the reality is AI won’t create as many new jobs as people assume. If AI does generate these anticipated productivity gains, then the benefits of this can’t flow exclusively to shareholders – a portion must reach the people it displaces. That legislation will be crucial for society and the future of work.
