The money chasing AI agents just got a shadow: three Asian giants betting nine figures that the real bottleneck isn't building agents, but keeping them from breaking everything.
The Summary
- Zenity raised $125 million from SoftBank, Hitachi, and LG to build security and governance tools for AI agents in enterprises
- CEO Ben Kliger says Asian enterprises are deploying agents across "every aspect of the enterprise," creating urgent need for oversight infrastructure
- The round signals a bet that agent governance, not agent capability, is the actual chokepoint for enterprise AI adoption
The Signal
SoftBank, Hitachi, and LG don't typically co-invest unless they see something their own operations need yesterday. This $125 million round into Zenity is less about financial upside and more about industrial necessity. These are companies already neck-deep in AI agent deployments. They're funding the seatbelt after they've already floored the accelerator.
The timing matters. Asian enterprises, particularly in manufacturing and consumer electronics, are ahead of their Western counterparts in agent deployment. Hitachi runs logistics networks. LG manages global supply chains. SoftBank backs dozens of companies automating everything from customer service to financial analysis. They all share the same problem: agents touching sensitive data, making autonomous decisions, and occasionally going sideways in ways that compliance teams can't audit.
"The real bottleneck isn't building agents. It's keeping them from breaking everything."
Zenity's pitch is governance infrastructure for the agent economy. Think of it as the enterprise IT policy layer, but for software that writes its own code and makes its own API calls. The platform monitors what agents access, flags unusual behavior, and creates audit trails that satisfy regulators who still think "AI governance" means a quarterly PowerPoint.
What makes this round interesting isn't the dollar amount. It's the customer list doubling as the cap table. When your Series C investors are also your design partners, you're not building for a hypothetical market. You're building for the specific pain points of companies already running hundreds of agents in production.
Key dynamics at play:
- Asian manufacturers deploying agents faster than Western peers, driven by labor cost pressures and supply chain complexity
- Regulatory environment in Asia-Pacific tightening around AI auditability, particularly in Japan and South Korea
- Enterprise IT teams realizing traditional security models don't work when the "user" is an LLM with API access to your entire stack
The agent security market is forming in real time. Traditional cybersecurity companies are still building for human threat models. Zenity is building for a world where the insider threat is your own autonomous workforce. The difference: human employees have predictable behavior patterns. Agents hallucinate, misinterpret instructions, and compound errors at machine speed.
The Implication
If you're deploying AI agents in an enterprise context, the question isn't whether you need governance tooling. It's whether you build it in-house or buy it from someone who's already stress-tested it at Hitachi scale. Most companies will buy. The cost of a governance failure, particularly in regulated industries, is orders of magnitude higher than a software licensing fee.
Watch for two things. First, Western enterprises following the playbook of these Asian investors and treating agent governance as critical infrastructure, not a nice-to-have. Second, regulatory bodies starting to mandate audit trails for autonomous AI systems. Zenity is positioning to be the default answer to "show us your agent compliance framework" before that question becomes mandatory.