Sentra, the agentic data security company, today announced that Capital One Ventures has led a strategic capital investment of $150 million in the company, joining existing investors Bessemer Venture Partners, Zeev Ventures, Standard Investments, Munich Re Ventures and Key1 Capital. It comes after a year of more than 300% growth, as enterprises race to put AI agents to work without losing control of their data.
Every company I talk to is under pressure to put AI to work. And nearly every security leader I talk to asks the same question: what can these AI agents actually reach?
Our answer is that you can't secure AI by chasing every agent. You have to follow your data.
The agentic era changes who touches your data
For decades, data security assumed people were the main consumers of data. People work at human speed, and their activity can be reviewed. AI agents don't fit that model. They run around the clock, pull information across systems and act on permissions that were set up for people. Gartner expects the average global Fortune 500 enterprise to go from fewer than 15 AI agents in 2025 to more than 150,000 by 2028.
That creates risk from two directions.
The first risk comes from inside. Most AI data exposure won't come from a rogue agent. It will come from an agent doing exactly what it was built to do, with access it never needed. Agents often inherit broad permissions from the accounts they run under, so they can reach sensitive data that has nothing to do with their task. Employees add to the problem every time they paste sensitive data into an AI tool the company hasn’t approved.
The second risk comes from outside. Attackers are moving faster too. They are exploiting critical flaws before fixes are available, often to steal the passwords and keys that unlock sensitive data. And AI agents now let a single attacker do what used to take a skilled team.
No company can patch every flaw in time or review every agent action. That is why security teams assume breach. The question then becomes what sensitive data can be reached, and by whom. It’s all about the impact!
AI security is a data problem
Much of AI security today focuses on the model or the prompt. Those controls matter. But an agent is only as risky as the data it can reach, which makes AI security a data security problem first.
This year's Zero Data Retention debate showed how much enterprises care. When AI providers changed what they kept, customers pushed back hard. But zero retention doesn't mean a provider never sees your data, and it doesn't decide what an agent can reach or what an employee sends. Only the enterprise can control that.
That is why we tell customers to follow their data rather than chase their agents. New agents launch faster than any team can catalog them, but they all draw on the same data. Know where your sensitive data lives, understand who and what can access it, and remove unnecessary access before an agent uses it.
This is what Sentra does. We discover and classify sensitive data across cloud, SaaS, data warehouses, on-premises and AI environments. We map every person, service account and AI agent to the data it can reach, and right-size that access. We stop sensitive data from flowing into AI tools employees use. And we do it within the customer's own environment, so their data isn't copied into ours for protection.
Chaim Sanders, CISO at Lyft, adopted Sentra as the data security solution to support the organization's growth. According to Sanders: “Our teams at Lyft, want to move fast with AI, but we need to know what data our agents can reach before we turn them on. Sentra gives us that answer across our whole environment, so security helps the rollout instead of slowing it down”.
Securing the data layer helps companies adopt AI faster. When a security team can see what an agent will touch, it can approve that agent with confidence.
Why this investment matters
Leading enterprises, including many Fortune 500 companies, already rely on Sentra. Capital One leading this round carries extra weight. Capital One is one of the largest regulated financial institutions in the United States, and it operates at a scale where getting AI and data security right is essential.
We will use the funding to build out agentic data security end-to-end. That means broader visibility into agents, copilots, and models and the sensitive data they touch, continuous least privilege for human and non-human identities, protection of data before agents reach it, and faster containment when an agent behaves unexpectedly. We will also scale our field and channel teams and expand into new geographies. We will grow our support and go-to-market teams and expand in Europe, where requirements such as GDPR and the EU AI Act make controlling AI data access a priority.
Follow your data
Every major technology shift has changed how businesses use and protect their data. AI agents are different because they don't just access information. They can make decisions and act on it, often without a person reviewing every step.
I believe the companies that will get the most out of AI are the ones that understand what their agents can access and have the right controls in place before giving them that access.
That's the future we're building toward at Sentra, and this investment will help us get there faster.
To our customers, partners, investors, and the Sentra team, thank you for the trust and for believing in what we're building. We've made tremendous progress, and there's much more ahead.
You can't secure the agentic enterprise by chasing every agent. You have to follow your data.
Yoav Regev Co-founder and CEO, Sentra
Chaim Sanders, CISO at Lyft, adopted Sentra as the data security solution to support the organization's growth. According to Sanders: “Our teams at Lyft, want to move fast with AI, but we need to know what data our agents can reach before we turn them on. Sentra gives us that answer across our whole environment, so security helps the rollout instead of slowing it down”.
