The London-based payment technology company Paysend must "strive" to obtain the permission to use artificial intelligence tools.
CEO Ben Chisell told CNBC that he is “a very skeptical person” about the hype surrounding this technology, yet he still sees its value.
Chisell has worked at Amazon, eBay, and Starling Bank. He estimates that Paysend could now complete about 25% more work with a team of the same size, but he believes that achieving results is not as simple as just distributing the latest technology to everyone.
“I know that considering tokenmaxxing and AI in this world, this approach may sound very unremarkable and not particularly sexy, but I truly believe it is very, very important,” said Chisell.
On the contrary, he limits the use of the AI tools to the employees who perform best, and usually waits for them to prove what results they can achieve with these tools first. He says that in the wrong hands, the AI tools can exacerbate bad habits such as laziness and poor judgment.
His broader principle is to resist the mentality of “fear of missing out” and to focus on the customer's problems.
“I know that considering tokenmaxxing and AI in this world, this approach may sound very unremarkable and not particularly sexy, but I truly believe it is very, very important,” said Chisell.
Promote AI within the company.
From the legal industry to software engineering and then to the financial sector, the pressure to prove the value of AI is increasing, and as a result, many companies are choosing to put this technology into the hands of as many people as possible.
Lloyd's Banking Group, headquartered in London, has issued 40,000 Microsoft 365 Copilot licenses. The bank stated that 97% of the employees who possess these licenses are actively using them. Sage, a software company based in Newcastle, claimed that its AI coding tools have saved hundreds of thousands of hours of work time; meanwhile, Grant Thornton UK, a professional services firm based in London, encourages all its employees to use Anthropic's Claude as part of its £500 million ($660 million) AI investment.
However, these figures do not necessarily translate into growth on the income statement. A study released by the UK government in January found that among businesses using AI, three-quarters stated that AI improved employee productivity, but 77% said that revenue remained unchanged.
Similar to Chisell, Unily, a British enterprise software company, CEO Lokdeep Singh also considers AI to be a kind of “superpower”, but it is best to reserve it for high-performing employees only. He is also wary of companies distributing AI to everyone just to cut costs.
“I personally think you should use it to expand your ambitions, rather than to reduce the team size,” he told CNBC. “But the reality is that in many industries, due to the macro environment, people tend to think this way: ‘Well, can I cut costs? Can I lay off employees?’ The truth is probably somewhere in between.”
Measuring the value of AI
At Paysend, Chisell expects employees to demonstrate what a particular tool has actually achieved. New employees will have one month to test the AI tool and explain the results they have obtained.
He recently tested this policy with a newly appointed search engine optimization manager. The employee requested to use Claude, and after a month, he informed Chisell that he no longer needed it. He said that the only help Claude provided him was to enable him to write briefings more quickly. Therefore, instead of continuing to use the general AI tool, the employee suggested developing a dedicated content briefing agent, and he is currently working on advancing this project.
At software company CoreView, the approach is different. The company's CEO, Simon Azzopardi, told CNBC that CoreView was initially "very enthusiastic" about AI and promoted it widely among software engineers.
They first used AI to investigate issues, located faults by correlating logs, events, and data, and also used it to generate technical documentation, which helped new engineers get up to speed more quickly.
“You’ll feel more efficient, but in reality, you’re not,” says Azzopardi.
Nowadays, he says that developers are increasingly coordinating teams composed of agents represented by AI. These agents are responsible for back-end and front-end coding, quality assurance, as well as documentation work. Azzopardi mentions that what used to take weeks to complete now may only take a few days.
However, Azzopardi warns that if employees use AI to do things outside of their area of expertise, the work produced by AI may not necessarily be more productive. “You might feel more efficient, but in reality, it’s not,” he says.
Nevertheless, when completing the launch for a major client of the UK National Health Service ( NHS ), CoreView only needed to add one engineer while continuing to deliver other projects, he said. The company did not lay off employees but continued to recruit, using the additional capacity to take on more work.
Positions after automation
As AI takes on more and more responsibilities that were previously done by humans, the question arises: what will happen to those employees who used to perform these tasks? Chisell believes that if companies help their employees transition to new roles, rather than directly replacing them after their work is automated, the companies will develop better.
For example, customer service staff have been solving complex problems for many years – this is a skill that is very valuable throughout the entire enterprise, he said. As for Chisell, technology should never replace human responsibility.
“Involving people shouldn’t just mean copying them along,” he said. “It should mean that people take responsibility themselves.”
