Maximizing AI tokens at work: Paysend CEO limits AI access to top employees

Paysend, a payments technology company headquartered in London, needs to gain access to artificial intelligence tools.

CEO Ben Chisell told CNBC that he’s “very cynical” about the hype surrounding the technology, but still sees value in it.

Chisel, who previously worked at Amazon, eBay and Starling Bank, estimates that Paysend has enabled teams of the same size to deliver about 25% more engineering work, but argues that getting results isn’t as easy as giving everyone the latest technology.

Instead, he limits the use of AI tools to his highest-performing employees, often only after they have proven what they can do with them. If AI falls into the wrong hands, bad habits such as laziness and poor judgment could become even more prevalent, he said.

His broad rules are to resist “fear of missing out” and focus on customer problems.

“I know it sounds very unappealing and unappealing as an approach given the world of token maxing and AI, but I think this is really, really important,” Chisel said.

From legal fields to software engineering to finance, the pressure to prove the value of AI is increasing, and many companies are responding by making the technology available to as many people as possible.

London-based Lloyds Banking Group has installed 40,000 Microsoft 365 Copilot licenses, and the bank says 97% of licensed employees are actively using it. Newcastle-based software company Sage said its AI coding tools had saved hundreds of thousands of working hours, while London-based professional services firm Grant Thornton UK encouraged all its staff to use Anthropic’s Claude as part of a £500m ($660m) investment in AI.

However, these numbers do not necessarily translate to your bottom line. A UK government study released in January found that three-quarters of companies using AI said their employees were more productive, but 77% said their bottom line had remained the same.

Like Chisell, Lokdeep Singh, CEO of British enterprise software company Unily, believes that AI is a “superpower” that only the best employees can have. He’s also wary of companies handing it out to everyone as a cost-cutting measure.

“Personally, I believe that you should use it to expand your ambitions instead of making your team smaller,” he told CNBC. “But the reality is that in many industries, again, the macro environment is driving people to think, ‘Can we cut costs? Can we cut employees?'” The truth is probably somewhere in between. ”

At Paysend, Chisell wants his employees to demonstrate how the tools actually contribute. New hires have one month to try out AI tools and explain their results.

He recently tested that policy with a new search engine optimization reader who requested access to Claude. After a month, the employee told Chisel he no longer needed it. The only thing he found useful was being able to write briefs more quickly, he said. Instead of a general-purpose AI tool, that employee suggested building a specialized content brief agent, which they are currently working on.

At software company CoreView, the approach is different. London-based CEO Simon Azzopardi told CNBC that the company was initially “very enthusiastic” about AI and had begun a widespread deployment of software engineers. They first used AI to investigate problems by correlating logs, events, and data, create technical documentation, and hire new engineers faster.

He said developers are now increasingly coordinating teams of AI agents to handle tasks such as back-end and front-end coding, quality assurance, and documentation. Azzopardi said work that used to take weeks can now take days.

However, Azzopardi cautioned that the work generated by AI is not necessarily productive if employees use it to build things outside of their expertise. “It feels like we’re more efficient, but we’re not,” he said.

Still, when onboarding a major customer in the UK National Health Service, CoreView only needed to add one engineer while continuing to deliver other projects, he said. The company is continuing to hire, rather than cutting jobs, and is using its extra capacity to take on more jobs.

As AI begins to take on more of humans’ existing responsibilities, the question is, what will happen to the employees who were performing these tasks? Chisel believes companies will grow by helping employees move into new roles, rather than replacing them after their jobs are automated.

For example, customer service representatives spend years solving complex problems, a valuable skill across the company, he said. For Chisel, technology should not replace human responsibility.

“Human involvement should not mean humans are CC’d,” he says. “It’s supposed to mean human ownership.”