Bay Exits Stealth With $37M Series A to Secure Enterprise AI Agents

Bay raised a $37 million Series A led by Battery Ventures, the Tel Aviv-based startup announced on November 10, 2026, bringing its total funding to $50 million as it emerged from stealth with a platform designed to control what enterprise AI agents can do.

The round is the hard evidence that a new category is forming: security tools that police AI agents from the inside, in real time, rather than blocking them at the network edge. Enterprises are handing agents legitimate permissions to touch cloud infrastructure, SaaS applications and employee devices, and the risk is no longer just whether an agent is compromised. It's whether an authorized agent takes a dangerous route to a legitimate goal.

What does Bay actually sell?

The founding story explains the product. Bay's founders watched a coding agent attempt to expose a cloud environment to the internet while completing a test. The agent had not been compromised. It was using legitimate permissions it had been granted.

That distinction shapes everything Bay built. The platform doesn't just ask whether an agent has permission to attempt an action. It asks whether the action is safe in context, assessing each proposed move against the full agent session.

Co-founder and CEO Yuval Zacharia described the objective as "enabling AI adoption at scale without losing control."

The mechanism works in milliseconds. Bay says its platform evaluates each proposed action in the context of the complete agent session and decides locally — in milliseconds — whether it can proceed. It learns from an organization's own agent activity and interprets real-time context, which lets it distinguish actions that look identical in isolation but carry different risk.

A few concrete capabilities define the platform:

• Session-level evaluation. Each action is judged against everything the agent has done in the session, not in isolation.
• Adaptive policies. The rules adjust as teams add agents and workflows, rather than requiring manual rule maintenance for every new tool.
• Model-agnostic coverage. The platform works across the models, assistants and agent frameworks an organization chooses.
• Integrated remediation. It combines AI discovery and posture management, session-level observability and real-time enforcement, with connections to existing security workflows for automated remediation.

Deployment is deliberately light. Bay says rolling the platform out across employee devices, cloud infrastructure and SaaS applications takes under five minutes and requires no additional persistent endpoint agent, kernel driver or network proxy.

The takeaway: Bay is betting that agent security has to happen at the session level, in the moment, or it doesn't work at all.

Who put money in?

The $37 million Series A was led by Battery Ventures. The total $50 million raised includes a previously undisclosed $13 million seed round co-led by YL Ventures and Striker.

The Series A also drew strategic names:

• Micron Ventures, the chipmaker's investment arm
• M12, Microsoft's venture fund
• Assaf Rappaport, CEO of cloud security firm Wiz, in a personal capacity

The investor list reads like a bet that agent security becomes the next layer of the enterprise security stack, the way endpoint and cloud security did before it.

Battery Ventures general partner Dharmesh Thakker said the decision to lead the round reflected Bay's technical approach and early enterprise customer wins.

In plain terms: the money follows both the product's architecture and the proof that real enterprises are paying for it.

Who is already using it?

Bay reports dozens of enterprise customers at launch, including Micron, Peloton, ClickUp and Wix. The company also says Fortune 100 companies used its platform to triple autonomous AI agent use — a claim that gets at the commercial pitch from the customer side. If agents are safer, companies let them do more.

Micron's participation is double-hatted: it is both a customer and an investor through Micron Ventures. Andrew Byrnes, Micron's director of venture capital, said the company had seen firsthand how Bay's technology could accelerate AI adoption while preserving the control and accountability required by a large enterprise.

That quote is the whole market case in one sentence — adoption speed and enterprise control are usually presented as a trade-off, and vendors in this space are selling the removal of that trade-off.

Who built it?

The founding team carries serious Israeli cybersecurity pedigree.

Zacharia founded Bay with chief product officer Aner Morag and research-and-development vice president Boris Danilovich. Zacharia previously led a 50-person department in Israel's Unit 8200, the military's storied signals intelligence unit, and later served as director of research and AI at Hunters.

Danilovich previously led a 90-engineer department at CrowdStrike. Morag joined Noname Security as an early employee and rose to vice president of product.

The pattern is familiar in Israeli security startups: alumni of Unit 8200 and major security vendors founding companies that attack an emerging threat category before the incumbents reorganize around it.

What happens next?

Bay plans to use the funding to expand its product and engineering teams and to grow sales, marketing and customer teams in the United States and Europe.

The round gives Bay the resources to scale, but the open question is technical as much as commercial. The company's central challenge is demonstrating that session-level enforcement can support wider agent autonomy without multiplying the rules security teams must maintain.

That's the test for the whole emerging agent-security category: if enforcing context-aware policies turns into managing thousands of bespoke rules, the product has just moved the operational burden rather than removed it. How Bay answers that will shape whether enterprise AI adoption actually accelerates on its watch.