Inside Citi's Stealth AI Revolution

For many major corporations, artificial intelligence remains confined to isolated pilot projects. Small teams experiment with tools, run tests, and present findings that rarely gain traction beyond a handful of departments. Citi has charted a different course. Over the past two years, the bank has focused on integrating AI into the daily workflow across the organization, moving it beyond the domain of a few specialists.

This initiative has cultivated an internal AI community of approximately 4,000 staff members, drawn from diverse functions including technology, operations, risk management, and client service. This figure, first reported by Business Insider, highlights how Citi developed its "AI Champions" and "AI Accelerators" programs to foster widespread participation rather than maintaining strict central oversight.

The degree of integration is significant. With a global workforce of about 182,000, Citi reports that over 70% of its employees now use company-approved AI tools in some capacity. This widespread adoption positions the bank ahead of many competitors who still limit AI access to technical teams or dedicated innovation labs.

From centralised pilots to team-level adoption

Citi's strategy began with people, not just tools. The bank invited employees to volunteer as AI Champions, providing them with specialised training, internal resources, and early access to approved AI systems. These individuals then supported their immediate colleagues, serving as accessible points of guidance rather than formal instructors.

This approach reflects a practical understanding of technology adoption. New tools often fail not due to a lack of capabilities, but because staff are unsure when or how to apply them. By embedding support directly within teams, Citi bridged the gap between experimental use and routine application.

Formal recognition was a key component. Employees could earn internal badges by completing courses or demonstrating how they used AI to enhance their specific tasks. While these badges didn't lead directly to promotions or salary increases, they helped build visibility and credibility for AI skills within the company. According to Business Insider, this peer-driven model accelerated adoption more effectively than traditional top-down directives.

Practical application with necessary safeguards

Citi's leadership has positioned this effort as a response to operational scale, not merely technological novelty. With vast operations in retail banking, investment services, compliance, and customer support, even minor efficiency improvements can yield substantial cumulative benefits. AI tools are now used to summarise lengthy documents, draft internal communications, analyse data, and assist in software coding. While these applications are not unique, the difference lies in their systematic deployment across the enterprise.

The focus on everyday utility also informs Citi's approach to risk management. The bank restricts employees to a set of vetted, company-approved tools, implementing clear guardrails on data usage and output handling. These constraints, while occasionally slowing experimentation, have increased management confidence in granting broader access. In heavily regulated sectors like finance, establishing trust in the process is often more critical than pursuing raw speed.

Lessons from Citi's approach to scaling AI

The architecture of Citi's program offers a key insight for other large organizations: successful AI integration doesn't require every employee to become an expert. It requires a critical mass of staff who understand the tools well enough to use them responsibly and guide their teammates. By training thousands instead of a select few, Citi reduced its dependence on a small, overburdened group of specialists.

There is a cultural dimension as well. Actively encouraging participation from non-technical roles sends a clear message that AI is not exclusive to engineers or data scientists. It becomes a standard component of modern work, akin to the adoption of spreadsheet or presentation software in prior decades.

This shift aligns with broader industry patterns. Research from firms like McKinsey indicates many companies struggle to transition AI projects from pilot to production, often citing skills shortages and unclear accountability. Citi's model mitigates some of these challenges by distributing practical ownership to teams while maintaining centralised governance and standards.

Naturally, the approach has its limitations. Peer-led adoption depends on sustained engagement, and progress can be uneven across different teams. There is also a risk that informal support networks become inconsistent, leading to disparities in benefits. Citi addresses this by periodically rotating AI Champions and continuously updating training materials as tools evolve.

A standout aspect is the bank's decision to treat AI as operational infrastructure rather than just an innovation project. Instead of framing the question as "Can AI transform our business?" Citi asked "Where can AI reduce friction in our existing work?" This pragmatic framing makes progress easier to quantify and alleviates the pressure to deliver sensational, immediate results.

The experience also challenges the common assumption that AI adoption must be exclusively driven from the top down. While Citi's senior leadership provided essential support, significant momentum came from employees who volunteered their own time to learn and share knowledge. In large, complex organizations, this bottom-up energy is difficult to manufacture, yet it is frequently the deciding factor in whether a new technology becomes embedded in the culture.

As more corporations move from testing phases to full-scale implementation, Citi's experiment provides a valuable case study. It demonstrates that scale is achieved not merely by acquiring more advanced tools, but by building people's confidence in using the tools they already have. For enterprises puzzled by slow AI progress, the solution may depend less on strategic presentations and more on supporting how work is accomplished, one team at a time.

See also: JPMorgan Chase treats AI spending as core infrastructure

Explore the intersection of AI and Big Data with industry leaders. The AI & Big Data Expo is happening in Amsterdam, California, and London. This comprehensive event is part of TechEx and is co-located with other leading technology events. Click here for more details.

AI News is brought to you by TechForge Media. Discover other upcoming enterprise technology events and webinars here.