Andreessen Horowitz, or a16z, said it has invested in AI agent security startup Runta. The firm did not disclose the size of the investment.
Runta says it wants to help companies manage and constrain AI agents in much the same way parents guide children. Founder Guanlan Dai previously worked on the technical team at Cloudflare and was a founding engineer at API connectivity startup Kong.
Dai said AI agents share some traits with growing children: they can carry out tasks on their own, but they still need boundaries, supervision, and permission controls. He said that just as parents put household safeguards in place and keep children away from credit cards, companies need to limit which important files AI agents can access, what actions they are allowed to perform, and how much money they can spend in a single operation.
Runta is developing what it calls an "AI agent guardianship" infrastructure layer for enterprises. The product is designed to manage agent permissions, security risks, and behavioral boundaries, with the aim of preventing data leaks, mistaken actions, or financial losses when autonomous AI systems execute tasks.
As more enterprises deploy AI agents with autonomous decision-making abilities, the need for trusted and secure agent management systems is becoming a new infrastructure demand. Runta wants to position itself as a parental-control layer for the AI agent era, offering identity management, permission controls, risk limits, and runtime oversight for enterprise agents.
The industry view cited in the report is that as AI agents move beyond simple assistant functions and become autonomous entities that can operate enterprise systems, process transactions, and carry out complex tasks, the market for security, governance, and compliance infrastructure around agents could see rapid growth.