The news: TransUnion announced an AI agent for its OneTru analytics platform that uses models from Google’s Gemini. The platform connects data sources that inform credit risk, marketing, and fraud management.

How it works: The AI Analytics Orchestrator Agent will let bankers query credit data and test credit scenarios. It’s designed to be explainable, in an effort to weather increasing scrutiny of AI models’ decision-making processes, meet banks’ model risk and governance standards, and comply with decades-old regulations that predate modern credit scoring models.

The agent breaks down its results, points to the computer code behind the output, and explains its logic to the user.

Zoom out: TransUnion’s agent is the latest example of large financial services companies and institutions finding new ways to query and analyze their vast troves of data. Experian, for example, pitches conversational AI as part of its data platforms. BNY said in December that it would integrate Google’s Gemini Enterprise into its in-house AI platform, letting employees build agents that use and act on the bank’s data. Goldman Sachs is working with Anthropic to build its own AI agents to accomplish myriad tasks across the bank.

Implications for banks: Financial institutions have an enduring data need and an equally large data problem. More than 40% of FIs globally said that data is critical to helping achieve business decisions, per our August 2025 report Unlocking Data and AI’s Promise in Banking.

But collecting crucial data points alone is not enough—unifying it poses a bigger challenge. AI tools like TransUnion’s only work when the underlying data is well organized and other models are already in place. Banks must ensure a sound data infrastructure before they implement agents.