Michael Bargury on building Zenity and securing the agentic AI frontier.
Zenity just raised $125 million to secure the era of agentic AI in the enterprise.
The rapid adoption of AI agents brings a new set of security challenges for organizations eager to innovate but concerned about how these new systems and AI tools might act inside the enterprise.
Three years ago, Zenity co-founders Ben Kliger (CEO) and Michael Bargury (CTO) were already working with some of the largest Fortune 50 companies on securing low-code and citizen development when generative AI changed the trajectory of the industry—and the company.
They made a bold decision: to follow what they were seeing with customers and refocus Zenity's mission on securing agentic AI.
That decision required immense conviction in their instincts. It also came down to a moment of clarity. During a visit to Microsoft, while Microsoft was shipping its first AI agent product, Michael flew to New York and spent the day walking around Central Park with Ben—five times. By the end of the day, they had decided to leave the old product behind and build for the future they were seeing.
UpWest co-led Zenity's seed round. Today, the company has more than 230 employees and protects Fortune 500 banks, hospitals, and pharmaceutical companies.
In this episode of Runtime, Shuly Galili sits down with Michael to unpack how Zenity went from that pivotal reset to securing what AI agents actually do inside the enterprise.
They discuss where conviction comes from, why "security of AI" is the wrong frame—and "security from AI" is the right one, how sharing a document can plant a false memory in someone's ChatGPT, why agent identity is table stakes and intent is the real problem, how enterprises can roll out new AI tools across an entire fleet in seven days, and why the security stack needs to move from build time to… runtime.
New episodes every other Wednesday.
Produced by POLDHU.
CHAPTERS
00:00 Intro: securing AI agents in the enterprise
03:26 SQL injection at 15, and the years at Microsoft
06:26 Starting Zenity on the unpaved road
07:39 2023: burning what they built
11:10 Security from AI, not security of AI
15:31 The Black Hat demo that hijacks ChatGPT
17:02 What CISOs are measuring wrong
20:58 How fast enterprises are actually moving
27:12 The hacker community, culture, and the team
36:34 Runtime, not build time
40:58 Advice to founders