Helena Almeida has watched the technology setting shift rapidly over the last few years. As an attorney and vice president managing counsel at Automatic Data Processing, she spent years dealing with the legal implications of artificial intelligence before the generative boom of late 2022. In 2025, the company expanded her remit, adding the title of chief AI legal officer to her existing responsibilities. Her focus now encompasses product compliance and the management of the company’s responsible AI initiatives, a shift that reflects the growing demand for oversight from corporate finance teams.
Almeida noted that the questions she receives have changed as the technology has matured. Initially, interest in AI tools was driven by the IT department, but the conversation has since moved to the executive level. Over the last two to three years, finance leaders and CFOs have become increasingly central to these discussions, bringing their own specific concerns to the table.
“It’s an interesting evolution,” Almeida said. “At first when people started really asking more questions about AI and the tools, it started with IT.” She pointed out that the current dialogue involves a mix of operational needs and regulatory compliance. The finance side of the house is particularly focused on how these tools integrate with existing financial controls and reporting structures.
Related: Lawmaker seeks AI hearings after Anthropic, OpenAI incidents
When asked what the most pressing questions from CFOs are, Almeida highlighted a specific concept she calls “human in the loop.” This phrase has become common in industry discussions, but Almeida argues that the execution matters more than the terminology. She explained that simply having a person click a mouse when prompted is not the standard she aims for.
“Human in the loop isn’t the same thing as meaningful human judgment,” Almeida said. “We all know that AI can look very polished and persuasive and everything, but the goal is where we’re designing a system that really encourages humans to say, ‘hey wait a minute’.” She emphasized that for payroll practitioners and finance professionals, oversight is not just a procedural box to check. It is a necessary step to ensure the output makes sense before it is acted upon.
Building trust at scale remains a significant challenge for companies integrating AI into their workflows. There is often a tension between moving quickly to adopt new technology and the caution required to ensure safety and accuracy. Almeida believes that ADP is handling this tension by prioritizing what she calls “the speed of trust.”
Related: Regulator flags risky fund return claims
“Part of the magic, and one of the things I’m very proud of that we’re doing at ADP is making sure that we’re developing at the speed of trust,” Almeida explained. She views responsible AI not as a constraint on innovation, but as the foundation that allows the company to scale its products safely. This approach involves pulling in subject matter experts from privacy, security, data, and ethics teams to review every new feature before it reaches the client.
ADP has introduced ADP assist, a suite of client-facing AI offerings designed to simplify complex administrative tasks. One of the capabilities within this umbrella involves analyzing payroll variances and identifying missing tax IDs. Almeida noted that these issues often arise during complex transitions, such as hiring a new worker or expanding into a new jurisdiction, where the variety of triggers can overwhelm standard processes.
The goal of these tools is to reduce the likelihood of human error. By flagging discrepancies early, the system helps finance teams avoid costly mistakes that could arise from overlooked details in payroll processing. This reliance on technology to handle the minutiae allows human professionals to focus on higher-level strategic oversight rather than data entry.
Related: Kenya needs smart rules for growth
For other companies looking to incorporate AI into their systems, Almeida offered a specific piece of advice regarding implementation. She suggested that finance leaders should not simply layer AI over existing processes. Instead, she recommended a holistic review of the full workflow to address the root causes of operational issues.
Her guidance extends to the composition of the teams driving these initiatives. Finance and HR leaders need to be at the table, she argued, even if they lack technical coding skills. Their focus on controls, accuracy, and the employee experience provides a necessary balance to the technical development of the tools.
These leaders should feel comfortable leaning in and voicing concerns about how the technology impacts their specific domains. Getting these practical considerations right is often what drives actual adoption within a company. When the technology aligns with real-world workflows and addresses genuine pain points, it becomes a tool that people actually want to use.
