Rethink Application Security
We’re building Sybil, an autonomous system of agents that thinks like an attacker. Sybil continuously probes live applications the way a real hacker would, validates what they find, and suggests the fix, keeping pace with how fast developers ship code. Teams close the gaps before anyone else finds them.
See Sybil in action
Automate hacker intuition
We help companies close security gaps as fast as they ship code.
Our leadership
.jpg)
Ari Herbert-Voss
- 1st security hire @ OpenAI
- Harvard ML PhD dropout
- Core research contributor to GPT-3 and Codex
.jpg)
Vlad Ionescu
- Red Teamer @ Mandiant, NCC Group
- Offensive Security Tech Lead @ Meta
- Founder of Red Team X
- Core Llama offensive research lead
Austin Zheutlin
- Former Chief of Staff/Head of Ops @ Trail of Bits
- Product/Operations at Series B/C Startups
- ex-Meta Election Integrity Ops
Built by researchers
Driven by conviction
Ari ran frontier research on large language models at OpenAI, and Vlad ran the red team at Meta. Both came away with the same conclusion: AI can hack. Paired with security engineers, it produces faster, higher-quality results than people working alone.
So they set out to redefine offensive security. The team they've built brings together AI research, internal and consulting red teaming, engineering at FAANG and leading self-driving car companies, and data science across the private and public sectors.
A culture built for builders
Our values shape how we work, what we build, and the kind of team we're proud to be. Pick one to run it.
We move with purpose, together. Nobody operates in a silo — not our team, agents, or customers.
We take the work seriously and ourselves lightly. The fun is functional. It gives us the confidence to try new things without fear. We mean it and we meme it.
We build deep expertise and share it generously, so our team and customers can leave every conversation more capable than when they came in.