Citi's HR Team Ran a Hackathon. An Internal Mobility Tool Won.
The News
Citi's HR department ran an internal AI hackathon in May, engaging approximately 180 HR employees across 35 teams over 10 working days. Led by Jeff Bienstock, Citi's head of HR AI, the initiative encouraged participants to "go crazy" with ambitious ideas, with risk, legal, and model governance teams brought in after ideation rather than before. The winning project combined employee profiles, résumés, skills data, job descriptions, and learning resources to recommend internal career opportunities — and Citi is already iterating on it to add learning recommendations that nudge employees toward development paths aligned with open roles. The hackathon also surfaced internal AI champions who now share use-cases at town halls and AMAs.
Why It Matters
The notable detail isn't the hackathon format — it's that a highly regulated financial institution chose to let HR staff experiment first and apply compliance controls second. Most TA teams in regulated industries never get past the "risk-averse starting point" Bienstock deliberately avoided, and that starting point produces risk-averse tools that nobody uses. The winning project is also telling: internal mobility won because it's the problem HR people actually live with daily. When you let practitioners identify their own problems rather than handing them a vendor's use-case menu, you get tools that solve real workflow pain.
Taking Action
All action is situated in your corporate context. These are starting points, not prescriptions — choose what fits.
Connect
Who to talk with in your enterprise
Your head of HR technology and your chief privacy officer or GC's office. The Citi model — ideate without constraints, then bring risk and legal in for scaling — is a specific governance choice. Ask both: could we run a structured experiment where TA staff prototype AI solutions to their own workflow problems, with a clear gate for compliance review before anything goes live? Your internal mobility lead — the winning project combined data sources (profiles, skills, job descriptions, learning resources) that probably already exist in your organisation but aren't integrated. That integration is a data architecture question, not an AI question.
Inspect
What to take a closer look at
Check your internal mobility metrics. What percentage of open roles in the last quarter were filled internally? If it's below 15%, you have the same problem Citi's hackathon winners were trying to solve. Also audit your AI governance process: does it gate ideation (preventing experimentation) or gate deployment (allowing experimentation with controlled scaling)? If your process prevents people from building prototypes, you're optimising for risk avoidance at the cost of adoption.
Reflect
What to think about more deeply
If Citi — a bank under more regulatory scrutiny than most organisations will ever face — let its HR team "go crazy" with AI ideation, what exactly is your governance process protecting you from?