At many large companies, artificial intelligence remains a side project. A small team tests the tool, runs a pilot, and presents results that are difficult to disseminate beyond several departments. Rather than taking a different path and limiting AI to experts, Citi has spent the past two years implementing the technology into day-to-day operations within the organization.
This effort brings the company’s AI workforce to approximately 4,000, with roles ranging from technology to operations, risk, and customer support. This figure was first reported by business insiderThis article details how Citi built its AI Champions and AI Accelerators programs to encourage participation rather than central control.
The scale of the integration is notable since Citi employs approximately 182,000 people worldwide, more than 70% of whom are currently using some form of company-approved AI tool, according to the same report. This level of utilization puts Citi ahead of many of its peers, which still limit access to AI to technology teams and innovation labs.
From central pilots to team-level implementation
Instead of starting with tools, Citi focused on people. The bank invited employees to volunteer as AI Champions, providing training, internal resources, and access to an approved early version of the AI system. Employees then served as local liaisons, rather than formal trainers, to support colleagues in their own teams.
This approach reflects a practical perspective on recruitment. New tools often fail not because they lack functionality, but because staff don’t know how and when to use them. By embedding support within the team, Citi reduced the gap between experimentation and day-to-day operations.
Training played a central role. Employees can earn internal badges by completing courses and demonstrating how they used AI to improve their tasks. Badges did not come with promotions or raises, but they helped increase visibility and credibility within the organization. According to business insiderthis peer-driven model helped AI spread faster than top-down commands.
Daily use, with guardrail
Citi leadership positions this initiative as a response to scale, not a novelty. For operations across retail banking, investment services, compliance, and customer support, small efficiency gains can quickly add up. AI tools are used to summarize documents, draft internal memos, analyze datasets, and assist with software development. These uses are not new per se, but the difference lies in how they are applied.
Our focus on day-to-day operations also shapes Citi’s attitude to risk. The bank limits employees to company-approved tools and has guardrails around what data they can use and how output is processed. This restriction slowed down some experiments, but allowed administrators to grant broader access. In regulated industries, trust is often more important than speed.
What Citi’s approach shows when it comes to scaling AI
The structure of Citi’s program suggests lessons for other large companies. AI implementation doesn’t require every employee to become an expert. There must be enough people to understand the tools well, apply them responsibly, and explain them to others. By training thousands instead of dozens, Citi has reduced its reliance on a small group of experts.
Cultural signals are also at play. Encouraging non-technical employees to participate sends the message that AI is not just for engineers and data scientists. It becomes part of the way we work, much like spreadsheets and presentation software were decades ago.
This change is consistent with broader industry trends. Research from companies like McKinsey and others shows that many companies struggle to move AI projects into production due to lack of talent and unclear ownership. Citi’s model avoids some of these problems by distributing ownership among teams while centralizing governance.
Still, this approach is not without its limitations. Peer-led adoption relies on continued attention, and not all teams will move at the same pace. There is also a risk that informal support networks will become uneven, with some groups benefiting more than others. City tried to address this issue by rotating champions and updating training content as tools changed.
What stands out is banks’ willingness to treat AI as infrastructure rather than innovation. Rather than asking whether AI can transform business, Citi asked whether AI can remove friction from existing operations. This framework makes it easier to measure progress and reduces the pressure to produce dramatic results.
This experience also challenges the common assumption that AI implementation must start at the top. Citi senior leadership supported the effort, but much of the momentum came from employees who volunteered their time to learn and teach. In large organizations, bottom-up energy can be difficult to generate, but it often determines whether new technology takes hold.
As more companies move from pilot to production, Citi’s experiment provides a useful case study. This shows that scale comes not by buying more tools, but by empowering people to use the tools they already have with confidence. For companies wondering why progress in AI feels slow, the answer may lie not in strategy materials, but in how teams actually work.
(Photo courtesy of Declan Sun)
See: JPMorgan Chase treats AI spending as core infrastructure
Want to learn more about AI and big data from industry leaders? Check out the AI & Big Data Expo in Amsterdam, California, and London. This comprehensive event is part of TechEx and co-located with other major technology events. Click here for more information.
AI News is brought to you by TechForge Media. Learn about other upcoming enterprise technology events and webinars.
