Israeli data security startup Sentra has raised $150 million to control what AI agents can reach inside corporate data systems. Led by Capital One Ventures, the round highlights a growing risk: automated identities, often granted overly broad permissions, read sensitive information continuously, frequently without any central oversight.
🔑 Key takeaways
- Sentra raised $150 million in a Series C led by Capital One Ventures; no valuation was disclosed.
- The platform maps access rights for users, service accounts and AI agents, including inherited permissions.
- Abnormal access attempts are flagged and can trigger automatic revocation of the relevant rights.
- The company reports growth above 300% over the past year, but the calculation basis is not specified.
A $150 Million Series C Led by Capital One Ventures
On November 4, 2026, Sentra Inc. confirmed a funding round of $150 million. Axios Pro describes it as a Series C, while the press release distributed by PR Newswire calls it a “strategic capital investment.” The sources do not explain the difference in terminology, which does not change the amount announced. Capital One Ventures, the venture arm of Capital One Financial Corp., leads the round and will take a seat on the board.
Existing investors Bessemer Venture Partners LP, Zeev Ventures, Standard Investments Inc., Munich Re Ventures and Key1 Capital also participated. Key1 had led a $50 million Series B in April 2025. According to SiliconANGLE and Ynet, total investment in Sentra since its founding now exceeds $250 million.
| Metric | Figure | Source |
|---|---|---|
| Round size | $150M, Series C | SiliconANGLE, Axios Pro |
| Previous round | $50M, Series B (April 2025) | SiliconANGLE, Ynet |
| Total raised since 2021 | More than $250M | SiliconANGLE, Ynet |
| Headcount | 130 employees, 65 in Tel Aviv | SiliconANGLE |
| Annual growth | More than 300%, basis not specified | SiliconANGLE, Ynet |
| Valuation | Not disclosed | No source |
AI Agents Are Outpacing Access Controls
Each agent deployed inside a company becomes an additional identity capable of reading data, often continuously and without a central control plane to supervise its activity. Sentra expects the number of agents deployed in large enterprises to rise sharply in the coming years, but believes only a small minority of organizations have the governance needed to precisely control what those agents can access. According to Ynet, some companies already use thousands of agents, and in some cases tens of thousands, while relatively few have full visibility over their permissions.
Inherited permissions and scattered data
Agents operate more autonomously than human users, access multiple systems and can receive broader permissions than their functions require. Sensitive data is also spread across clouds, software platforms and on-premises servers, which makes mapping it difficult for traditional tools. Axios sums up the challenge: “AI needs data, but companies are still struggling to map and secure that data.”
“AI agents are the most active consumers of enterprise data.”
Yoav Regev, Co-founder and CEO of Sentra
Regev argues that many agents inherit access to data that was never fully classified or governed. Traditional perimeter controls, designed for human users, were not built for agents operating at machine speed.
Securing the Data Instead of Filtering Every Request
Sentra’s approach does not involve reviewing every request an agent sends in real time. The company focuses first on securing the underlying information, following three steps:
- Map permissions and identify excessive privileges granted to users or automated systems.
- Classify sensitive information according to each client’s business activity.
- Eliminate obsolete or particularly risky data that no agent has a reason to access.
When an agent attempts to reach sensitive data outside its usual scope, the attempt is flagged. The platform can then automatically withdraw the permission or sharing setting in question, while the most serious cases are escalated to the person responsible for the dataset. After an incident, legal and compliance teams can obtain a record of everything the agent could have reached, along with the business context of each dataset.
Sentra says analysis takes place inside the client’s environment and that sensitive data never leaves it. The company argues that tools copying customer data into a vendor’s environment create a second data estate to protect, with added retention and residency risks for regulated sectors.
“Organizations can make sure the data these agents access is already secured.”
Yoav Regev, Co-founder and CEO of Sentra
What the Announcement Leaves Unanswered
Several elements remain unpublished. Sentra has not disclosed its valuation, and the sources cite no trading data, market reaction, or analyst or investor commentary. The growth figure of more than 300% over one year, reported by SiliconANGLE and Ynet, does not specify whether it refers to revenue, annual recurring revenue or another performance metric.
The split of proceeds between product development, commercial expansion and hiring has not been detailed, nor has any secondary share component. Sentra employs 130 people, 65 of them in Tel Aviv, and has offices in New York and Tel Aviv. Ynet reports that the company plans to strengthen its sales operations in Europe, the Middle East and Africa, expand its US federal business and invest in distribution channels and technology partnerships.
Founded in 2021, the company was started by four technology and cybersecurity veterans: Yoav Regev, former head of the cyber department of Israel’s Unit 8200; Asaf Kochan, former commander of that unit and Sentra’s chairman; Ron Reiter, chief technology officer and former head of Oracle’s Tel Aviv development center; and Yair Cohen, vice president of product, previously at Microsoft and Datadog.
Outlook and Scenarios
Governance of non-human identities could become a distinct category within enterprise cybersecurity. Three scenarios stand out. In the base case, large organizations gradually integrate classification and access mapping into their AI programs, supporting demand for this type of platform. In a more cautious scenario, agent deployments slow for lack of adequate governance frameworks, which would limit commercial expansion. In a disruptive scenario, a major incident involving a poorly controlled agent would accelerate purchases of such tools, benefiting vendors already in position.
The core point, as Regev frames it, is no longer only to keep unauthorized people out of corporate information, but to ensure organizations remain in control of their data when automated systems can retrieve and process it independently. The coming quarters will show whether this thesis translates into verifiable revenue and durable adoption.
Sources
Cet article est publié à titre informatif et éducatif. Il ne constitue en aucun cas un conseil en investissement. Faites vos propres recherches (DYOR) avant toute décision.
