Capital One Software is expanding its Databolt data security platform with new capabilities aimed at making unstructured data safer to use in analytics and AI initiatives. The company announced the updates at RSAC 2026 in San Francisco.

The headline feature is a set of upcoming Databolt workflows designed to inspect unstructured formats, including PDFs, emails, transcripts and office documents, then identify and protect personally identifiable information. Capital One Software said the goal is to apply protections such as tokenization in a way that preserves semantic structure, so data remains useful to large language models while reducing exposure.

“Our goal is to unlock the value of sensitive data by making security seamless across the data lifecycle,” Prashant Prahlad, SVP and head of product at Capital One Software, said in a prepared statement. “With these new innovations, Databolt will help organizations move past the security bottlenecks that act as a barrier to innovation.”

Alongside the unstructured-data work, Capital One Software outlined additional platform updates. Those include a fully self-hosted deployment option for regulated organizations, enhanced protection policies with granular access controls, and a token analytics dashboard intended to give both technical and non-technical users visibility into system health and throughput.

The company also described “domains,” a way to logically segment configurations inside a single tenant, and said future releases will expand connectivity for collecting contextual metadata across complex data stacks.

Cync Software, a Capital One Software customer, positioned Databolt as a way to embed data-centric controls into the data layer. “With Databolt, we are reinforcing a data-centric security approach by embedding security directly into our data layer, reducing risk exposure across our lending and financial platforms,” said Prasad Alapati, CEO and president of Cync Software.