We build open-source governance and safety infrastructure for autonomous AI agents, from Røyksund, Norway.
- super-tanks — open-source (Apache 2.0) governance layer that makes autonomous AI agents safe in production. 10 enforcement layers mapped to the OWASP Top 10 for Agentic Applications (2026) and the EU AI Act. Currently under independent security audit.
- supertanks-guard-mcp — local MCP server with human-in-the-loop guardrails for AI agents: code/text scanning, deny-by-default action gating, and a tamper-evident audit log.
- go-gate — minimal human-approval gate for agent actions.