Sabrina Halper Show · 3 min · Oct 7, 2026

Wiz shifted to AI before models were fully capable

YC

Yinon Costica

Co-founder, Wiz

Co-founded Wiz, the cloud security company Google bought for $32 billion, and leads its product organization.

Sabrina HalperHost · Host, Sabrina Halper Show
Has worked as an investor at HOF Capital and founded SHS Media.

About this segment

Costica says Wiz started experimenting with assistive AI early, then made a broader shift after anticipating where models were heading. The team organized work around red, blue, and green agents within months.

From “Google’s $32B Bet on Cybersecurity in the AI vs. AI Era | Wiz Co-Founder Yinon Costica”, Oct 7, 2026. Watch the full episode

“So we started experimenting. We first did the assistive technology from very early on”

YCYinon Costicaat 45:29 · Sabrina Halper Show45:29

Takeaways

Early models were weak, but the team chose to gain hands-on experience.
The broader shift began about a year and a half before the interview.
Red, blue, and green agents became the working framework.

Watch 2:57 from 44:27

In this episode

Transcript

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

44:33
that would have been very hard to achieve otherwise. And when you guys started, you obviously were seeing the explosion of the cloud in a way. And you were like, we need to go focus on this. No one else is. Maybe six months later, ChatGPT launched in 2023. And oh no, years later, ChatGPT launched. What was the moment where you were like, we actually need to put a bunch of the company's resources, talent, and hiring towards just focusing on this additional sort of AI problem. Yeah. So early days, the models weren't as powerful as today, right? So we looked at it and we said, but it's not really powerful to change the way we do things. It doesn't do things for me at the quality we need. But we realized that we, you know, in the end, even if it's a very, even if it's a very long train or slow train,
45:28
it doesn't matter. We have to jump on the one of the, you just have to be on the train, right? So we started experimenting. We first did the assistive technology from very early on, right? We said, okay, we need to create the hands-on experience. So we will be latching into the train whether it will move very slow or move very fast over the years you know the months in that case we realized that the models were progressing very fast and we could do more and more and i think that a year ago we we started doing the major shift more than a year and a half ago so a year and a half ago we looked at the the threat landscape let's say half a year after gpt launched and we were basically saying okay if we we fast forward this this is where things will get to these are the investments we have to do right away once we had this realization we started by
46:30
talking about the mental model and the framework once we upgraded the mental model and we said okay this is a new framework we have to be thinking through in order to solve ai problem it was a very rapid transformation and I know it sounds crazy but it was like three four months we were already seeing the capabilities we call them the red agent for pen testing blue agent for defense a green agent for remediation so we created this agent declare that actually implemented a the foundation for where we are today so yeah and very first the other side of it is securing AI. That's something that we did continuously because from our perspective, we secure cloud and everything within them. AI became yet another thing that we just have to secure. So that came