Bay raises $50M to police AI agents, exits stealth in 2026

Lead

Tel Aviv-based Bay left stealth on November 10, 2026, nine months after founding, with $50M raised across two rounds. The $37M Series A was led by Battery Ventures, with Micron Ventures participating. Micron is also a paying customer. The earlier $13M seed was co-led by YL Ventures and Striker Venture Partners. Bay did not disclose its valuation.

The company sells software that sits between an AI agent and the systems it touches. It judges each proposed action against the task the agent was given and its prior behavior, then permits it, blocks it, or sends it to a human for approval.

What Does Bay Actually Do?

Bay inspects agent actions before they execute, rather than analyzing logs afterward. A component called Bay Enforcer returns a verdict in milliseconds, according to the company's product description. Bay reports a false positive rate below 1%, a figure that has not been independently verified.

A scanner named Bay Wave inventories the agents, extensions and local models running on employee machines. That addresses a basic problem: many security teams do not know which agents their staff have installed. The software deploys through existing mobile device management or endpoint detection tools in under five minutes.

A third piece, Lighthouse AI, studies 30 days of agent activity and drafts security policies from how teams actually use the tools. The aim is to spare security staff from writing rules for behavior they cannot yet predict.

Who Is Behind Bay?

Bay was founded by three veterans of the Israeli military intelligence unit 8200. Yuval Zacharia is CEO and previously ran research and AI at security firm Hunters. Aner Morag is chief product officer, and Boris Danilovich leads research and development. Danilovich's former research team at CrowdStrike largely moved over to Bay, giving the startup an unusually experienced bench for a nine-month-old company.

The company employs about 35 people, mostly in Israel. Angel investors include Wiz CEO Assaf Rappaport, Cisco chief product officer Jeetu Patel and former Zscaler security chief Manoj Apte.

Who Is Already Using It?

Named customers are Micron, Wix, ClickUp and Peloton, alongside dozens of organizations in finance, insurance, technology, industry and media. Bay says several Fortune 100 companies run the platform and that those customers tripled their use of autonomous agents while keeping existing security policies intact. That is a company claim, and it comes without disclosed baselines or customer names for the Fortune 100 group.

The Micron link deserves a note. A strategic investor that also buys the product signals real deployment, but it also means one backer has a commercial interest in the startup's reference story.

Why Did Investors Move So Fast?

Investors moved quickly because enterprises are deploying agents faster than security teams can govern them. A $37M Series A for a company under a year old is large by historical standards. It follows a $13M seed that was itself sizable for a pre-product team, and the step-up suggests the seed backers saw early customer traction rather than a slide deck.

The pattern fits the Israeli cyber playbook of elite military-unit founders, a crowded but well-funded category, and rapid enterprise pilots. Agent security is attracting entrants from several directions, including endpoint vendors, identity companies and other startups. Bay's bet is that action-level control, not discovery or monitoring alone, becomes the durable layer.

Skepticism is warranted on two fronts. Large platform vendors can bundle agent controls into existing suites, which compresses pricing for point products. And a verdict engine that blocks legitimate work will be switched off by frustrated users, so the sub-1% false positive claim will be tested at scale.

What Comes Next for Bay and Agent Security?

Bay will likely spend the new capital on hiring and on expanding beyond its first enterprise accounts. The company's stated goal, in Zacharia's words, is for security to become "the reason AI ships faster, not the reason it stalls." That framing positions the product as an enabler of adoption, which is how many security buyers now justify budget.

Several outcomes are plausible. Bay could grow into a standalone platform if agent usage keeps expanding across departments. It could instead become an acquisition target for a larger security vendor that wants runtime agent controls, a path well-worn by Israeli cyber startups. A third scenario is that model providers and cloud platforms build comparable guardrails natively, narrowing the independent market.

Outlook

Bay has raised $50M in nine months on the claim that AI agents need a real-time gatekeeper. Early customers and a strategic investor give the pitch credibility, though valuation, revenue and independent accuracy data remain undisclosed. The next test is whether its enforcement holds up across thousands of employees without slowing them down.