Paysend CEO Ben Chisell limits AI tool access to his top-performing employees, arguing that broad rollouts risk amplifying poor judgment — even as rivals like Lloyds Banking Group deploy tools to tens of thousands of workers.
While many companies are racing to distribute artificial intelligence tools across entire workforces, Ben Chisell, CEO of London-headquartered payments technology company Paysend, is taking a deliberately narrow approach — restricting access to his highest-performing employees only.
Chisell told CNBC he is a "very cynical man" about the hype surrounding AI, even as he acknowledges the technology's value when deployed carefully.
Paysend, which focuses on cross-border payments, now delivers roughly 25% more engineering output with the same headcount, Chisell said. But he attributes that gain not to broad adoption, but to selective deployment.
"I know it sounds very unglamorous and unsexy as an approach given the world of tokenmaxxing and AI, but I think it's really, really important," Chisell said.
His core concern is that AI, in the wrong hands, magnifies poor judgment and laziness rather than correcting for them. His general rule is to resist "fear of missing out" and keep the focus on solving customer problems.
New hires at Paysend receive a month to experiment with AI tools and must explain what the tools actually contributed. Chisell recently applied that policy to a new search engine optimization lead who had requested access to Anthropic's Claude. After a month, the employee told Chisell he no longer needed the general-purpose tool — the main use he had found was producing briefs more quickly. Instead, the employee proposed building a dedicated content-brief agent, which he is now developing.
The approach stands in contrast to companywide rollouts underway at several large U.K. firms. Lloyds Banking Group has distributed 40,000 Microsoft 365 Copilot licenses, with 97% of employees who hold one actively using it. Newcastle-headquartered software company Sage reported its AI coding tools saved hundreds of thousands of work hours. London-based professional services firm Grant Thornton UK has encouraged all staff to use Claude as part of a £500 million ($660 million) investment in AI.
Those broad deployments have not consistently moved the bottom line, however. U.K. government research published in January found that 75% of AI-using businesses reported improved workforce productivity, but 77% reported no revenue change — a gap that underscores the difficulty of translating efficiency gains into financial results.
Lokdeep Singh, CEO of British enterprise software company Unily, echoes Chisell's philosophy, describing AI as a "superpower" best reserved for high performers. He also cautioned against using broad AI rollouts primarily as a cost-cutting mechanism. "I personally believe you should use it to expand your ambition rather than make the team smaller," Singh told CNBC. "But the reality is that in many industries, again, because of the macro environment, it is being thought as: 'OK, can I cut costs? Can I cut employees?' The truth is probably somewhere in between."
London-based CoreView CEO Simon Azzopardi described a more aggressive initial rollout — the company was "very gung-ho" on AI, distributing it widely among software engineers for tasks including log correlation, technical documentation, and onboarding. Engineers now increasingly coordinate teams of AI agents handling back-end and front-end coding, quality assurance, and documentation, with work that previously took weeks now taking days.
Still, Azzopardi flagged a hidden productivity trap: AI-generated output is not necessarily useful when employees use it to build things outside their areas of expertise. "You feel more efficient, but you actually aren't," he said.
For Chisell, the human element remains non-negotiable regardless of how capable AI tools become. He believes companies that help employees transition into new roles — rather than replacing them as their tasks become automated — will be better positioned long term. Customer-service workers, for example, carry years of problem-solving experience that is transferable across a business.
"Having a human in the loop shouldn't mean human CC'd," Chisell said. "It should mean human ownership."
As pressure mounts on companies to demonstrate returns on AI investment, Paysend's restrained model offers an early data point in an ongoing debate over whether broad or targeted deployment better converts the technology's capabilities into durable business value.
Sara Montes de Oca is the Editor in Chief of TechEchelon. Previously a correspondent and producer in Washington, D.C., covering business, finance, and politics.
More from Sara →Director of National Intelligence Jay Clayton warned Friday that AI-related foreign espionage is a reality for both U.S. companies and the government, pointing to the federal case against a former OpenAI employee accused of installing a recording device in an executive conference room.
Amazon unveiled the $75 Bee Frontier wearable and $100 Bee Eclipse smartphone-mounted recorder on Thursday, the first hardware update to its Bee AI listening device lineup since acquiring the San Francisco startup in July 2025.
Apple, Amazon, and Microsoft each unveiled new AI-powered consumer devices within a single week, making the 2026 holiday season the first major commercial test of mainstream demand for generative AI hardware.
Written for readers who already know the basics — markets, AI, and the policy decisions that shape both.