Agentic AI arrived with an awkward side effect. Every assistant a company switches on behaves like a new employee with login rights, reading databases around the clock, and few enterprises have a central map of what those digital workers may touch.
Sentra, an Israeli data security vendor founded in 2021, has raised $150M to sell exactly that map. Capital One’s venture arm wrote the Series C cheque and takes a board seat, with Bessemer, Key1 Capital, Munich Re Ventures, Standard Investments and Zeev Ventures tagging along. Sentra has now pulled in more than $250M since launch without disclosing a valuation. Business growth topped 300% over the past year.
Its approach begins with knowing what sits where. Every dataset gets a label tied to the business that owns it, and beside that comes a record of who can open it, from named staff and shared machine logins to the models themselves, nested rights included.
The behavioural layer is the interesting part. When an agent reaches for material outside its normal pattern, the system raises a flag, can revoke access or sharing on its own, and escalates the gravest cases to the owner. It also hands legal teams a timeline of everything the agent could have opened, annotated with each file’s business meaning.
Analysis stays within the buyer’s own cloud. Sentra argues that rivals who copy data back to their servers hand customers a second estate to defend, a residency headache in regulated sectors. Stripping out stale or risky records and capping service accounts at the minimum also shrink the surface an agent can wander into.
Funding goes toward visibility across models, faster containment and new geographies. Chief executive Yoav Regev, who built the firm with Asaf Kochan, Ron Reiter and Yair Cohen, sees agents as the busiest readers of corporate records.
