AI-Native Data Security, Governance, and Classification

Bedrock Data was founded in 2022 by Pranava Adduri and Ganesha Shanmuganathan, and launched publicly in March 2024 with a $10M seed round led by Greylock. Adduri was a founding engineer at Box and Rubrik and grew a product line past $200M at AWS; Shanmuganathan came from Cohesity and VMware with more than fifty patents. The company is based in San Jose, California. Its argument is that conventional data security posture management stops at finding sensitive data, which leaves teams with a long list of findings and no way to tell which ones an attacker or an over-permissioned AI agent could actually reach.

The platform discovers and classifies data across cloud, SaaS, and on-premises stores using AI-driven classification that reads business context rather than pattern-matching alone. Findings feed the Metadata Lake, a graph that joins data to identity at petabyte scale and holds sensitivity, lineage, entitlements, and access together as one source of truth. Combining precise classification with live access context is what lets the platform rank real exposure ahead of theoretical exposure, which is the company's answer to the false-positive volume that makes many DSPM deployments hard to act on.

Two architectural choices follow from that. The scanning layer is serverless and avoids brute-force full scans, which is how Bedrock Data claims full visibility without the cost curve that usually accompanies it. And analysis runs in place: data is not copied out of the customer's environment to be examined. On top of the discovery layer sit governance controls aimed squarely at AI, covering which agents and LLMs may reach which data and what they are permitted to do with it once they have it. Customers are concentrated in financial services, healthcare, biotech, and technology. The company was previously named Bedrock Security and its old domain now redirects to bedrockdata.ai.