Israeli artificial intelligence security startup Bay AI Inc. launched today with $50 million in funding to let enterprises vet each action an AI agent tries to take before it runs.

Bay provides software that tracks the agents working across a company, whether they sit on an employee’s laptop, inside a cloud workload or in a software-as-a-service application. One set of policies covers them all.

First-generation AI security tools mostly filter prompts and responses. Bay argues that misses where the damage happens, since a harmless-looking prompt can still end with an agent deleting a production database.

The founders ran into that kind of failure themselves when one of their own coding agents, working with its normal permissions, tried to expose a cloud environment to the internet to finish a test. They still have the screenshot.

Bay installs through the mobile device management or endpoint detection and response tools a customer already has, and the company says it takes under five minutes. A scanner called Wave inventories the agents, extensions and local models on each machine and leaves nothing running once it finishes.

Each time an agent proposes an action, the AI tool running it hands the request to Bay Enforcer, a static binary, through a native hook. The check happens on the machine itself with the whole session as context, and Bay said a verdict comes back in milliseconds, sometimes as a prompt asking the user to approve the step. Anything judged too risky never runs.

The policy behind those verdicts comes from Bay’s Lighthouse system, which drafts a separate one for each team from the last 30 days of agent activity, so security staff are not writing rules every time someone picks up a new tool. Teams can test a policy against the previous month first and see what it would have stopped. Bay said false positives run under 1%.

Security leaders are often handed an enterprise-wide AI rollout with a year-end deadline, co-founder and Chief Executive Yuval Zacharia said, or find that agents have already spread across the company and “it’s the Wild West.” She wants security to become “the reason AI ships faster, not the reason it stalls.”

Micron Technology Inc., Wix.com Ltd., ClickUp Inc. and Peloton Interactive Inc. are among the dozens of enterprise customers signed in the nine months since Bay was founded. Fortune 100 companies on the platform tripled their use of autonomous agents in just under 30 days, the company said.

Zacharia founded Bay with Chief Product Officer Aner Morag and Vice President of Research and Development Boris Danilovich. The three knew each other from the Israel Defense Forces’ Unit 8200, where Zacharia led a department of about 50 people before becoming the first intelligence officer there to cross into offensive cyber operations. After the military she ran security research and AI at Cyber Hunters Ltd.

Danilovich was an early employee at Preempt Security Inc. and stayed six years at CrowdStrike Holdings Inc. after it bought the startup in 2020. Most of Bay’s research team followed him over. Morag was the first employee at Noname Security.

The $50 million comes from two rounds. Battery Ventures LP led the larger $37 million Series A round, and Micron’s venture arm, Micron Ventures, also participated after the chipmaker started out as a Bay customer. Wiz Inc. Chief Executive Assaf Rappaport put in money personally. The remaining $13 million came from a previously undisclosed seed round co-led by YL Ventures Ltd. and Striker Venture Partners.

Yoav Leitersdorf, managing partner at YL Ventures, said his firm looked at a crowded field of AI endpoint security companies before picking Bay. He believes it can become “as fundamental to enterprise AI adoption as the network firewall was to the internet era.”

Engineering and product hiring will continue in Tel Aviv, where Bay is based, with the sales, marketing and customer teams growing in the U.S. and Europe. Enterprises can buy the platform now.

Photo: Bay