Antigen
ActiveContinuous offensive security for the enterprise.
About
Antigen is an AI adversary for the enterprise. It maps, exploits, and secures your attack surface through nightly automated offensive testing and monthly human red team assessments. Every finding is delivered with traces, reproduction steps, and remediation directly in GitHub, Linear, or the Antigen Platform. We work with companies in financial services, healthcare, retail, and government where security is trust. We're backed by Lightspeed, DST Global, and Susa Ventures.
Founders · 2
co-founder @ antigen. prev. ml infra at tesla autopilot. law school dropout. Before Zag, Saad trained reasoning models with Ashish Vaswani, author of the famous Attention Is All You Need paper. At Nuro, he developed an ML Compiler framework that enabled running the perception layer across multiple GPU's. And at Tesla he built infrastructure that halved the time it took to train Planning & Controls models. He also served at the Supreme Court of Pakistan and dropped out of UChicago Law.
Co-Founder @ Antigen. Previously product at Google Ads and ML at Google Search. Previously at Google, shipped product features for new advertisers and built large-scale ML infrastructure for measuring cross-site conversions on Search. Took a year off college to join as second employee at a YC startup. Separately scaled a bootstrap SaaS to $30k MRR. Also wrote low-level imaging software for silicon wafer manufacturing.
Launch
Custom review agents that build, test, and run your code before it lands.
Zag builds and runs AI review agents that execute in real environments on every PR. Developers describe tasks in TypeScript or Swift; Zag provisions dedicated Apple Silicon in sandboxed sessions to build, test, run, and verify apps, including iOS/macOS targets, before code lands. The service aims to replace diff-only reviews with validated, runnable code checks.
Formerly “Claybird”, “zag.dev”, “Zag”
Related startups
Autonomous offensive and defensive security platform
Agentic Offensive Security at Scale
