Bay Exits Stealth With $50 Million to Secure Enterprise AI Agents

AI security startup Bay has emerged from stealth with $50 million in total funding, including a $37 million Series A led by Battery Ventures. The round also pulls in Micron Ventures, M12 (Microsoft's venture fund), and Assaf Rappaport, CEO of Wiz, alongside a previously undisclosed $13 million seed led by YL Ventures and Striker Venture Partners.

Founded in 2026 and headquartered in Tel Aviv, Bay has already signed dozens of enterprise customers, including Micron, Peloton, ClickUp, and Wix. That customer traction at such an early stage is rare, and Battery Ventures is leaning on it as proof the technology works.

What problem does Bay actually solve?

Autonomous AI agents differ from traditional software in one critical way: they independently select actions and tools to complete assigned objectives rather than following narrowly defined instructions. That flexibility boosts productivity, but it becomes a liability when agents hold access to sensitive information or critical systems.

Bay's founders did not arrive at this problem through market research. They watched a coding agent attempt to expose a cloud environment to the public internet while completing a legitimate testing assignment. The agent was not compromised and had received no malicious instruction. It simply chose an unsafe operation while pursuing an authorized goal.

The takeaway: autonomous systems can create security risk even when operating entirely within legitimate permissions. For enterprises rolling out agents across departments, that gap sits between productivity and liability.

How does the platform work?

Bay's platform evaluates agent behavior before potentially risky actions execute. It analyzes proposed actions in the context of the agent's broader session and checks them against organizational policies. The company says these decisions happen locally, within milliseconds.

The system does not require administrators to write individual rules for every agent and tool. Instead, it uses observed activity to interpret behavior and adapt enforcement. Bay says the technology distinguishes between actions that look similar in isolation but carry different risks depending on context.

Coverage spans employee devices, cloud infrastructure, and software-as-a-service applications. The platform bundles five capabilities:

• AI discovery
• Security posture management
• Session observability
• Adaptive guardrails
• Real-time enforcement

Bay claims initial deployment takes less than five minutes, with no persistent endpoint agent, kernel driver, or network proxy. The platform also integrates with existing security workflows for remediation and incident response, and it works across different AI models, assistants, coding agents, and agent development frameworks. That model-agnostic design lets organizations set common controls without betting on a single AI provider.

Who is behind the company?

Bay was founded by CEO Yuval Zacharia, Chief Product Officer Aner Morag, and VP of R&D Boris Danilovich. Zacharia held cybersecurity and intelligence leadership roles in Israel's Unit 8200 and served as Director of Research and AI at Hunters. Danilovich previously led roughly 90 engineers at CrowdStrike. Morag was an early employee of Noname Security and rose to Vice President of Product.

What do the investors claim?

Dharmesh Thakker, General Partner at Battery Ventures, framed the bet plainly: "In Bay, we saw a company we felt had a differentiated technical approach that was already winning enterprise customers at an early stage. That combination gave us conviction to lead the round. We believe Bay can become a core control layer for companies putting AI agents to work at scale."

Yoav Leitersdorf, Managing Partner at YL Ventures, made a bigger claim: "We evaluated a crowded field of companies addressing AI agent security, and Bay stood out because it can understand and control individual agent actions, from the endpoint to the cloud, while fitting into infrastructure enterprises already use. We believe Bay can become as fundamental to enterprise AI adoption as the network firewall was to the internet era."

Andrew Byrnes, Director of Venture Capital at Micron, added: "We believe that this trust layer will become essential infrastructure for the agentic era."

Investor enthusiasm, of course, is the cheapest commodity in a funding announcement. The firewall comparison in particular sets a high bar that no early-stage company has cleared yet.

Does security actually accelerate AI adoption?

Zacharia argues security is the enabler, not the obstacle: "Bay was built to be the one platform that secures every way the enterprise uses AI, so security becomes the reason AI ships faster, not the reason it stalls."

Bay reports that enterprise customers using its platform have increased autonomous agent adoption by as much as threefold. That figure is self-reported, and the company did not disclose how it was measured or over what period. The claim deserves scrutiny, but it aligns with Bay's core thesis: effective controls reduce fear of unauthorized activity and unlock wider deployment.

The company also did not disclose revenue, valuation, or customer contract sizes — standard omissions in stealth exits, but gaps nonetheless.

What comes next?

The $50 million will fund expansion of Bay's engineering and product teams and growth of commercial operations in the United States and Europe. The company is betting that as agents spread across business applications, development environments, cloud infrastructure, and enterprise systems, demand for runtime control of agent actions will grow with them. If Zacharia's team is right, securing agents mid-execution becomes the prerequisite for the agentic enterprise.