Bay’s founders got their idea the uncomfortable way. While testing an AI coding agent, they watched it try to expose a cloud environment to the public internet. Nobody had told it to. It was acting within its permissions. On November 10 the Tel Aviv company came out of stealth with $50 million to stop that kind of thing in production.
The money is a $37 million Series A led by Battery Ventures plus a $13 million seed. Bay was founded in early 2026, has about 35 employees and already lists dozens of enterprise customers, including Micron, Peloton, ClickUp and Wix. CEO Yuval Zacharia founded it with chief product officer Aner Morag and R&D head Boris Danilovich.
Permissions aren’t the problem
Most security tooling asks whether an identity is allowed to do something. Agents break that model. An agent with legitimate access to a cloud console can still make a terrible decision with it. Bay watches what the agent is trying to do, evaluates each proposed action against the whole session, and decides locally, in milliseconds, whether it can go ahead.
The platform combines discovery of the agents running in a company, posture management, session-level observability and real-time enforcement. It plugs into existing security workflows so violations trigger remediation automatically. Bay says its false-positive rate is under 1%.
Security as an accelerator
The headline claim is that Fortune 100 customers tripled their use of autonomous agents within about 30 days of deploying Bay. That’s the pitch to buyers: security teams block agent rollouts because they can’t see or control what agents do, and a control layer lets them say yes.
YL Ventures’ Yoav Leitersdorf, an investor, compares it to the network firewall in the internet era. That’s ambitious. But the problem is real, and every company handing agents real permissions will need something in this slot. Bay plans to expand engineering in Tel Aviv and build sales and customer teams in the US.