At many large companies, artificial intelligence still operates on the margins. Small teams run pilots, test tools, and share results that rarely move beyond innovation units. Citigroup has taken a different approach, quietly weaving AI into daily work across the organisation rather than keeping it confined to specialists.
Over the past two years, the bank has built an internal AI network of around 4,000 employees drawn from technology, operations, risk, and customer-facing roles. The initiative, first reported by Business Insider, reflects a deliberate shift away from centralised experimentation toward broad, practical adoption.
Citi employs roughly 182,000 people worldwide, and more than 70% now use firm-approved AI tools in some capacity. That level of participation puts the bank ahead of many peers that continue to limit AI access to technical teams or innovation labs.
From pilots to people
Instead of starting with software, Citi focused on building internal capability. Employees were invited to volunteer as “AI Champions,” gaining access to training, internal resources, and early versions of approved tools. Their role was not to act as formal trainers, but to support colleagues within their own teams as local points of reference.
The idea was simple: new technology often stalls not because it lacks potential, but because people are unsure how to apply it to real work. By embedding support inside teams, Citi shortened the distance between experimentation and routine use.
Training became a key driver. Employees could earn internal badges by completing courses or demonstrating how AI improved their own workflows. While the badges did not come with financial rewards or promotions, they helped build visibility and credibility. According to Business Insider, this peer-led model helped AI adoption spread more quickly than top-down mandates.
Everyday use, carefully managed
Citi’s leadership has framed the rollout as a response to scale rather than novelty. With operations spanning retail banking, investment services, compliance, and customer support, even small efficiency gains can have a meaningful impact.
Employees use AI tools to summarise documents, draft internal communications, analyse datasets, and assist with software development. None of these applications are new on their own. What sets Citi apart is how consistently they are embedded into everyday tasks.
That focus also shapes the bank’s risk posture. Staff are limited to firm-approved tools, with clear guardrails around data usage and output handling. While these constraints can slow experimentation, they have helped build trust among managers and compliance teams. In regulated industries, that trust often determines whether technology adoption can scale.
Lessons for large organisations
Citi’s experience highlights a broader lesson for enterprises struggling to move AI beyond pilots. Widespread adoption does not require every employee to become an expert. It requires enough people to understand the tools well enough to use them responsibly and explain them to others.
By training thousands rather than dozens, Citi reduced reliance on a small group of specialists. It also sent a clear cultural signal: AI is not reserved for engineers or data scientists. It is becoming part of how work gets done, much like spreadsheets or presentation software did in earlier decades.
That approach aligns with wider industry research. Surveys from firms such as McKinsey consistently show that many companies fail to move AI projects into production due to talent gaps and unclear ownership. Citi’s model addresses both by distributing ownership within teams while keeping governance centralised.
The strategy is not without challenges. Peer-led adoption depends on sustained engagement, and progress can vary across departments. Citi has responded by rotating Champions and updating training as tools evolve.
What stands out most is the bank’s decision to treat AI as infrastructure rather than innovation. Instead of asking whether AI could transform the business, Citi focused on where it could reduce friction in existing work. That framing makes progress easier to measure and lowers the pressure to deliver dramatic, short-term results.
A quiet shift with lasting impact
As more companies look to move AI from experimentation to execution, Citi’s approach offers a practical case study. It suggests that scale comes less from acquiring new tools and more from helping people feel confident using the ones already available.
For organisations wondering why AI progress feels slow, the answer may lie not in strategy decks or pilot programs, but in the everyday habits of teams and the support they receive to change how work gets done.
