For most companies, Artificial Intelligence is stuck in “pilot purgatory.” You run a test, the results look promising, but the technology never leaves the IT department. It becomes a shiny toy for the specialists, while the rest of the company continues business as usual.
Citi has effectively flipped this script.
Instead of hoarding AI access within innovation labs, the bank has spent the last two years aggressively decentralizing it. The result? A massive internal workforce of roughly 4,000 “AI Champions” and a staggering 70% adoption rate across their 182,000 global employees.
Here is why their approach is working—and what founders need to steal from their playbook.
1. Influence Needs to Flow Sideways, Not Down
The biggest bottleneck in tech adoption isn’t software; it’s human behavior. When management sends a memo saying “Use this tool,” compliance is often grudging. When a colleague at the next desk says, “Hey, this tool saved me two hours yesterday,” curiosity takes over.
Citi capitalized on this by building a peer-to-peer network. They didn’t just hire external trainers; they upskilled volunteers from customer support, risk, and operations. These employees became the local experts—the “AI Champions”—who could translate technical capability into practical workflow improvements for their specific teams.
2. Gamification Without the Gimmicks
To identify these champions, Citi introduced a system of internal badging. Importantly, these badges weren’t tied to immediate pay raises or promotions. They were tied to status and visibility.
By allowing employees to earn credibility by showcasing how they used AI to solve actual problems, the bank tapped into intrinsic motivation. This generated a “bottom-up” energy that is notoriously difficult to manufacture in large organizations.
3. Make It “Boring” Infrastructure
Perhaps the most critical insight for business owners is how Citi framed the technology. They stopped asking, “How can AI transform our entire business model?” and started asking, “Where is the friction?”
The tools are being used for mundane tasks: summarizing documents, analyzing datasets, and drafting notes. By treating AI as core infrastructure—like Excel or email—rather than a radical innovation, they lowered the barrier to entry.
The Takeaway for Founders
If you are waiting for a perfect strategy deck before rolling out AI, you are already behind. Citi’s execution proves that scale doesn’t come from buying more expensive tools. It comes from demystifying the ones you have.
The lesson is simple: Stop treating AI like a science project for your engineers. Put it in the hands of your operators, give them safe guardrails, and let them figure out how to work faster.
