While major financial institutions and software providers race to distribute generative artificial intelligence to every employee on their payroll, London-based payments company Paysend is requiring workers to earn their access. According to an October 17 report by CNBC, Paysend CEO Ben Chisell limits general-purpose AI tools to high-performing staff members who can demonstrate exactly how the software contributes to their daily output.

Chisell, a former executive at Amazon, eBay, and Starling Bank, described himself as cynical about the broad industry hype surrounding enterprise AI, even as he estimates that Paysend now delivers roughly 25% more engineering work with the same-sized team. Rather than handing out licenses across the company, Paysend gives new hires a one-month trial period to experiment with AI tools and present what they have accomplished before granting ongoing access.

Earned Access Versus Blanket Rollouts

Paysend’s selective policy stands in sharp contrast to the high-volume deployment strategies favored by many global enterprises. Across the financial, legal, and technology sectors, executives have spent millions of dollars putting conversational assistants and coding tools into as many hands as possible.

In the United Kingdom alone, Lloyds Banking Group has distributed 40,000 Microsoft 365 Copilot licenses, reporting that 97% of recipients actively use the software. Newcastle-based software provider Sage says its AI coding tools have saved hundreds of thousands of work hours, while professional services firm Grant Thornton UK has encouraged its entire staff to adopt Anthropic’s Claude as part of a £500 million ($660 million) AI investment.

Global hardware and consulting giants have followed a similar path. According to The Economic Times, electronics manufacturer TDK reported at the Bloomberg New Economy Forum that more than 40% of its team members use AI alongside more than 2,000 deployed AI agents. At the same forum, PwC leadership noted the firm gave its entire workforce access to AI tools roughly 18 months ago—despite high deployment costs—to encourage hands-on learning and reduce fears of job displacement. Similarly, Fujitsu Global reported that approximately 100,000 employees, representing nearly its entire workforce, were using AI as of June 2026 following a companywide rollout started in 2025.

At Paysend, however, Chisell warns that unrestricted access to general-purpose models can amplify poor judgment or lazy work habits. His one-month trial policy recently prompted a newly hired search engine optimization lead to give up access to Anthropic’s Claude. After testing the chatbot for a month, the employee told Chisell that its only clear benefit was generating content briefs faster. Instead of keeping the general-purpose license, the employee proposed building a dedicated content-brief agent tailored specifically to that task.

The Gap Between Adoption and Revenue

The debate over how widely to distribute AI licenses highlights a growing challenge for corporate leadership: high adoption rates do not automatically produce financial returns. U.K. government research published in January found that while three-quarters of businesses using AI reported improved workforce productivity, 77% stated that their revenue had not changed.

Industry leaders gathered at the New York Stock Exchange earlier this month echoed those concerns. According to Fortune, speakers at the Fortune AIQ Summit warned that handing out software licenses without redesigning underlying workflows often fails to produce real organizational maturity.

Diana David, director of futures at ServiceNow, told summit attendees that having 95% of a workforce equipped with AI tools does not translate into productivity without proper training and operational discipline. According to ServiceNow’s Enterprise AI Maturity Index, 57% of top-performing organizations invest in ongoing AI upskilling, compared with just 4% of other companies. The gap is even wider in recruiting and retaining specialized AI talent, where 68% of leading firms invest heavily compared to 10% of their peers.

Paul Fipps, president of global customer operations at ServiceNow, emphasized that measurable gains come from rebuilding business processes rather than simply automating existing habits. Fipps pointed to Standard Chartered, where targeted AI deployment across employee-facing functions has deflected 77% of internal staff requests. Meanwhile, other software firms have found success by focusing initial rollouts on specific technical bottlenecks. Simon Azzopardi, CEO of software company CoreView, told CNBC his firm initially gave AI broadly to its software engineers to correlate system logs, investigate bugs, write technical documentation, and speed up onboarding.

What Selective AI Policies Mean for Workers and Leaders

For enterprise executives and tech workers, Paysend’s experiment offers a pragmatic alternative to unstructured software rollouts. Requiring employees to justify their tool usage forces teams to identify specific operational bottlenecks—such as building a specialized workflow agent rather than relying on an open-ended chatbot subscription.

Chisell also argues that organizations should use automation to move staff into higher-value roles rather than cutting headcount. He pointed to customer-service representatives who have spent years resolving complex client disputes as prime candidates to take on broader operational responsibilities once routine inquiries are automated. In his view, keeping a human in the loop must mean genuine human ownership and accountability, not merely copying an employee on an automated decision.

How are Paysend’s AI restrictions evaluated and enforced?

Public reporting offers differing descriptions of how formalized the policy is. While CNBC reports that Chisell limits general-purpose AI tools to high performers and gives new hires a one-month trial to demonstrate value, Traders Union characterizes the access as tied to performance metrics. Neither report details a standardized, companywide grading rubric or independent verification for Chisell’s estimate of a 25% increase in engineering output.

Did Paysend restrict AI access due to cybersecurity or data privacy risks?

No. While business leaders at the Bloomberg New Economy Forum emphasized that information security, risk management, and governance are major factors when scaling enterprise AI, Chisell’s stated rationale at Paysend centers strictly on employee judgment, work quality, and proving practical business value rather than preventing data leaks.

Sources

Read More