Sentra, a data security company focused on what autonomous agents can reach, has raised a $150 million Series C round led by Capital One Ventures. The company says the round follows more than 300% year-over-year growth, and it targets a problem many enterprises have not solved: AI agents now read and write corporate data continuously, often with permissions nobody has reviewed.
Sentra described the round as a strategic capital investment. Existing investors Bessemer Venture Partners, Zeev Ventures, Standard Investments, Munich Re Ventures and Key1 Capital also took part, and a Capital One Ventures representative will join the company's board. According to BankInfoSecurity, Sentra was founded in 2021, has about 210 employees and has raised roughly $253 million across four rounds.
Agents Are Now the Heaviest Data Users
Sentra's chief executive, Yoav Regev, put the shift bluntly in the company's release: "AI agents are now the busiest consumers of enterprise data." The release argues that many agents inherit access to data that has never been classified or governed. Each agent is effectively a new identity that touches data at machine speed, and there is no central control plane that sees all of those touches at once.
The company says only a small minority of organizations report having governance in place to control what their agents can access. That is Sentra's own characterization, and it has not been independently verified, but it matches the gap that security teams describe as they move agents from pilots into production.
What Sentra Is Building
Sentra's approach is to secure data at rest before an agent reaches it, rather than inspecting every agent request in real time. Regev describes an "access intelligence layer" that maps people and non-human identities, including AI agents, to the data they can touch. In his telling, that mapping could support time-limited permissions, so an agent receives only the access a specific task requires.
Regev also framed the core rule plainly: "You don't want to take your sensitive data and push it to the models." Sentra currently helps organizations find sensitive data and reduce the chance that it reaches the wrong AI tools or the wrong users through copilots.
The company said it will use the new capital for:
- Agentic data security, including visibility into agents, copilots and the models they touch
- Continuous least-privilege enforcement for human and non-human identities
- Protecting data before agents reach it, and rapidly containing agents that behave unexpectedly
- Broadening its data loss prevention products into a wider data security platform
- Expanding sales, marketing and new geographic markets
What Enterprise Security Teams Should Take From This
The practical lesson is that an AI agent should be inventoried and governed like any other identity. Teams should know which agents exist, which systems they can reach and which data those systems hold. Access should be granted per task rather than standing, and sensitive data should be classified before it is exposed to an agent's context window.
Several details remain unannounced. The release does not give pricing, customer counts, release timelines for the new capabilities or technical detail on how the access controls work. Buyers should ask for those specifics during evaluation rather than relying on the category label.
Why It Matters
Investors are betting that agent security will become a separate budget line, not a feature inside existing identity or data tools. Capital One Ventures leading the round signals interest from a financial company that runs large data estates and has direct exposure to regulated information.
The bigger question is whether data-centric controls can keep pace with agents that change what they do from one task to the next. Sentra's answer is to shrink the damage before an agent acts. Whether that approach holds up as agents grow more autonomous will be one of the defining tests for the category over the next two years.
