Two launched products, one focus: trustworthy AI.
Super Tanks — open-source AI-agent governance
Most AI security reacts after something goes wrong. Super Tanks is built to prevent it: ten enforcement layers mediate every action an autonomous agent takes before it reaches a tool, a model, or the outside world — so a single bypassed layer is not the end.
- Prevents, not just detects — the controls sit before the action, not after the incident
- Mapped to the OWASP Top 10 for Agentic Applications and aligned to the EU AI Act
- Open source (Apache 2.0) · runs fully offline on local models · published System Card & threat model
The ten enforcement layers
Every agent action passes through all ten — defense in depth, so one bypassed layer is not the end.
Lovefilter
A free Chrome extension that uses AI to hide online hate and harassment, helping people keep their social feeds civil. Live on the Chrome Web Store — you can install it today.
How we work
We build in the open. Super Tanks is Apache-2.0 licensed, its System Card and threat model are public, and the same architecture runs two AI agents in everyday production — so the failure modes we learn from are real, not hypothetical.
