Companies create AI committees to guide decisions
Bruno Ferreira
Rogério Vieira/Valor
Companies across different sectors are establishing governance structures to set priorities and organize artificial intelligence projects and training. According to a Gartner study of 1,800 global leaders, more than half—55%—worked at organizations that had maintained AI committees since 2024. Experts expect the trend to gain momentum in the coming years as companies seek to oversee all activities involving new language models.
“The goal is not to restrict innovation, but to create conditions for it to happen at scale, safely and predictably,” said Maikon Silva, an IT risk partner at Grant Thornton Brasil. “The organizations that capture value in this area are precisely those with clear mechanisms for decision-making, accountability and monitoring.”
At retailer C&A, the AI committee began with six members and now has eight, including executives involved in the company’s digital strategy and outside experts such as retail consultant Alberto Serrentino.
“The meetings focus on identifying trends, tracking the progress of projects and assessing where AI can have an impact,” explained Bruno Ferreira, C&A’s technology and digital transformation director. The company has more than 15,000 employees and 339 stores. “We also discuss which initiatives should advance first, which investments make sense and how to ensure the technology addresses business challenges.”
Among the projects the committee discussed, Ferreira said, is a personal-shopper system whose progress the group has tracked since the early stages of development. In practice, the application “understands” consumers’ preferences, purchase history and intent to provide more relevant recommendations, he explained.
The greatest benefit of maintaining a committee is that it provides strategic direction on a rapidly evolving subject, Ferreira noted. “The committee helps share knowledge and ensure responsible adoption,” he said. It also brings together areas such as technology, data and business, making AI a corporate issue rather than solely a technological debate. “We are already seeing improvements in alignment and in identifying opportunities across different areas of the company.”
At Einstein Hospital Israelita, AI is governed not by a council but through an institutional process involving bodies that evaluate, guide and monitor solutions, said Igohr Schultz, executive director of digital transformation. “These include a multidisciplinary governance committee that analyzes initiatives involving data, software and AI,” he noted.
According to Schultz, the governance structure began taking shape in 2021, before generative models became widely popular. It gained momentum as new projects emerged across the healthcare system, which has 33,800 employees. “It became clear that we needed a better-organized way to assess solutions from the design stage, considering not only technical aspects but also the context of use, data and the risks involved,” he explained.
Schultz, who did not disclose the profiles or number of professionals assigned to the initiative, said the governance committee’s responsibilities include evaluating projects developed internally and products offered by the market. The analysis considers responsible-use principles such as security, privacy and human oversight.
“Governance plays a guiding role, supporting teams from the project-creation stage and helping them incorporate technical, ethical and regulatory safeguards,” he said. “At the end of each assessment, formal recommendations are issued to guide the next steps.”
One application the team reviewed was a system supporting care delivery. Schultz recalled that the discussion began while the system was still being designed, before its integration with existing programs. “Addressing it early allowed us to define the scope of use, the type of information to be used, and how performance would be monitored.”
Vivian Broge, vice president of human relations and marketing at technology company Totvs, believes organizations should establish an AI council once AI systems in the workplace become a pillar of business strategy. “That turning point occurs when project complexity requires moving beyond isolated initiatives to centralized governance that can provide scale and alignment with corporate compliance.”
Last year, Totvs, which has 12,000 employees, created a program to coordinate and advance AI teams and projects under a governance model. “It is through this program that we centralize decisions,” she explained. “It consolidates a journey of initiatives developed over many years.”
The program brings together more than 400 professionals. Of those, 78% have worked at the company for 3 to 18 years, while 16% hold leadership positions. Participants range from technical specialists, including data scientists and software architects, to representatives of business-support areas such as legal affairs and customer relations. A project management office (PMO) and a management committee support the initiative, and the committee reports to Totvs’s steering committee.
“In practice, this structure both decides and executes—from the technical development of the product portfolio to data governance and team training,” said Broge, who participates in the initiative.
According to Silva of Grant Thornton Brasil, one of the main advantages of organizing an AI council is that it allows good ideas to move forward. “Effective governance accelerates AI adoption because it replaces the approval cycle that often leaves projects stalled between departments such as compliance and IT,” he said. “It can also establish who is accountable when something goes wrong.”
The committees should include the IT director and business leaders who serve on a rotating basis and represent those using AI, not only those approving it, Silva suggested. “This composition becomes even more important as investments increase and require return metrics.”
Translation: Todd Harkin
