Sentra's $150M Series C Bets Data Security Must Govern AI Agents Before They Reach the Store

Sentra has closed a $150 million Series C led by Capital One Ventures, positioning its platform as agentic data security for enterprises that are rolling out copilots and autonomous agents faster than they can classify what those agents can touch. The November 4, 2026 company release (New York and Tel Aviv) says Capital One Ventures led a strategic investment joining existing backers Bessemer Venture Partners, Zeev Ventures, Standard Investments, Munich Re Ventures, and Key1 Capital, with a Capital One representative joining Sentra’s board. Sentra reports more than 300% year-over-year growth. Co-founder and CTO Ron Reiter’s blog post the same day frames the capital as fuel to scale engineering and go-to-market across North America and Europe, not as an end state.

The product thesis is blunt: data security was built for people, and AI agents are now among the busiest consumers of enterprise data. Sentra argues three pressures hit at once—business demand for AI, sensitive data sprawled across cloud, SaaS, and on-prem, and non-human identities multiplying without a central control plane. Rather than inspecting every agent request in flight as a pure perimeter problem, Sentra emphasizes protecting data at rest where it lives, then governing which human and AI identities can reach it.

From DSPM point tool to agent-aware control plane

Sentra’s own history, as told in the CTO post, started as cloud Data Security Posture Management for AWS, Azure, and GCP visibility, then widened into a unified data security platform covering IaaS and PaaS stores, Microsoft 365 and Google Workspace–class SaaS, on-prem fileshares, and endpoints. The security domains the company stacks on that discovery engine are Data Access Governance (mapping identities—including AI agents—to data and enforcing least privilege), Data Loss Prevention (stopping sensitive leakage into public environments or generative AI apps), and hygiene/remediation (shadow data, redundant files, exposed secrets). Classification remains the foundation: Sentra says it combines small language models with contextual intelligence so policies are not written against mislabeled buckets.

The PR narrative breaks agent risk into before, during, and after. Before deployment, Sentra aims to shrink exposure—find and classify sensitive data, map inherited access for people, service accounts, and agents, enforce least privilege for non-human identities, and remove high-risk data agents should never see. During an incident, data-aware detection is meant to flag unusual sensitive access and automate remediation such as revoking access or risky sharing. Afterward, the platform is pitched as evidence of what an agent could have reached, with business context for legal and compliance response. Architecture detail the company stresses for regulated buyers: scanning runs inside the customer environment; sensitive data stays put; only metadata leaves. That “assume breach” framing also shows up in the CTO post’s discussion of secrets exposure and blast-radius governance when an account or agent identity is compromised.

Agent identities put data-access controls under pressure

Sentra is making an infrastructure bet on agent identity. CEO Yoav Regev’s line is that traditional perimeter controls were not built for agents moving at machine speed, so the practical control is securing the data those agents reach and deciding which identities may touch it. Lyft CISO Chaim Sanders is quoted saying teams want to move fast with AI but need to know what agents can reach before turning them on. Founders listed are Yoav Regev, Asaf Kochan, Ron Reiter, and Yair Cohen (2021). Advisors named are H-F & Co. for Sentra and Wachtell, Lipton, Rosen & Katz for Capital One Ventures.

Growth and product breadth are company-reported; peer-review claims in the release are not reproduced here as scored tables. The round also funds field, channel, brand, and geography. The durable technical claim remains specific: unify DSPM, detection/response, access governance, and DLP around data-at-rest classification that treats AI agents as first-class identities, without copying the customer’s estate into a second vendor cloud. Sentra’s Series C prices the bet that agentic adoption will keep outrunning manual classification unless discovery, identity mapping, and containment sit on the same platform.