Capital One executive Maureen Jules-Perez is calling for enterprise AI systems that demand fewer prompts, fewer interfaces, and less workflow effort from employees. Her view places human judgment at the center of workplace automation, rather than treating AI as another tool workers must constantly manage.
The approach addresses a growing concern inside large organizations. AI may speed up individual tasks, yet poorly designed systems can add steps, split attention, and force employees to learn new interfaces. Jules-Perez argues that effective AI should remove that friction and give people more time to apply their knowledge.
Reducing the Work Around AI
Many workplace AI products rely on employees to write detailed prompts, review generated material, and move results between separate systems. Each task can reduce the time saved through automation.
Jules-Perez’s position suggests that success should not be measured only by how well a model responds. Companies must also consider how much effort workers spend getting that response and putting it to use.
Effective enterprise AI should “reduce prompts, interfaces and workflow friction” while giving employees more room for judgment and expertise, Jules-Perez said.
That principle favors AI built into existing work processes. Instead of asking employees to open a separate assistant and explain each request, a system could use approved business context to support routine steps.
The goal is not simply fewer clicks. It is a lower mental burden. Employees should be able to focus on decisions, exceptions, and customer needs rather than operating the technology itself.
Human Expertise Remains Central
Jules-Perez’s comments also draw a line between assistance and replacement. Enterprise AI can prepare information or handle repeated work, but employees still bring context, accountability, and professional judgment.
This distinction matters in banking. Financial institutions manage sensitive data, regulatory duties, fraud risks, and decisions that can directly affect customers. AI output may be useful, but it must fit within controls and clear lines of responsibility.
A practical enterprise system should support several goals:
- Remove repeated steps instead of creating new ones.
- Present useful information inside established workflows.
- Leave important decisions with qualified employees.
- Make review and accountability clear.
This model can also support employee trust. Workers are more likely to use AI when it helps them complete a task without hiding how the result will influence a decision.
A Different Measure of AI Value
Companies often assess AI through speed, output quality, or adoption. Jules-Perez’s argument adds another measure: how little attention the tool demands from its user.
That standard could change how organizations plan AI projects. A technically capable product may offer limited value if it requires constant prompting or sends employees through several screens. A narrower system may deliver greater gains if it works quietly within a familiar process.
There are trade-offs. More automated systems require reliable data, careful permissions, and testing. Reducing user input must not remove needed oversight. Companies also need ways for employees to question results, correct errors, and handle unusual cases.
What Businesses Should Watch
The next phase of enterprise AI will likely be judged by work outcomes rather than demonstrations. Leaders will need to track time saved, error rates, employee experience, and the quality of decisions made with AI support.
Jules-Perez’s message offers a pragmatic test for future investment. If an AI system adds another interface, more instructions, or extra review without clear gains, it may be shifting work instead of reducing it.
The stronger model places technology in the background and expertise in the foreground. For Capital One and other large employers, the key question is no longer whether employees can use AI. It is whether AI can support employees without getting in their way.
