# XOR

> Cybersecurity

- Website: https://www.xor.tech/
- LinkedIn: https://www.linkedin.com/company/xor-tech/

XOR produces cyber reinforcement learning training data to train AI models in cybersecurity. The company provides RL environments, verified task trajectories and evals that AI teams use to train coding agents to autonomously find and fix their own security flaws.

- First investment: Pre-Seed
- Founders: [Tobias Heldt](https://www.linkedin.com/in/tobias-heldt-214561264/), [Henk Birkholz](https://www.linkedin.com/in/iot-ng/)
- Locations: Munich, San Francisco
- In our portfolio under: Cyber & TrustTech (AI that fixes CVEs)

## In the press

- [XOR Advances IETF Protocol for Verifiable Agent Conversations](https://www.linkedin.com/posts/xor-tech_a-judge-in-connecticut-sanctioned-a-self-represented-activity-7496582278727389184-3BlN?utm_source=share&utm_medium=member_desktop&rcm=ACoAAEu5OaQBahO9aKZujhtWrMxDV_VmhIbMNQQ), XOR, Aug 2026

[The whole portfolio](https://www.maetch.vc/index.md)

---

This is the markdown version of https://www.maetch.vc/portfolio/xor, written for agents. All pages: https://www.maetch.vc/llms.txt
