Bay AI launches with $50 million to vet AI agent actions
The AMW Read
Bay adds a funded entrant focused on agent action controls, with initial enterprise customers but unverified performance claims.
Bay AI launches with $50 million to vet AI agent actions
Israeli security startup Bay AI has launched software that checks proposed AI agent actions before they run. The company raised $50 million across two rounds: a $37 million Series A led by Battery Ventures and a $13 million seed round co-led by YL Ventures and Striker Venture Partners. Bay says Micron, Wix, ClickUp and Peloton are among dozens of enterprise customers.
Bay is entering the market for controls around autonomous agents, where a safe-looking prompt can still lead to a risky action. Its Enforcer tool checks requests locally through hooks in AI tools, using the session as context. Bay says its Lighthouse system drafts team-specific policies from 30 days of activity and lets security teams test those policies against the same period. The company reports false positives below 1%, though the article does not describe how that figure was measured.
For enterprise buyers and investors, the test is whether action-level checks can prevent harmful changes without interrupting useful work. Bay says its scanner inventories agents, extensions and local models, while its software can also track agents in cloud workloads and software-as-a-service applications. Buyers can use the policy replay feature to assess what rules would have blocked before enforcement. The company also says some decisions prompt a user for approval; it has not provided independent evidence in the article of how its controls perform across customer environments.
