Research

Applied research.

We turn frontier AI research into a concrete edge. Our own open-source projects, backed by numbers and scientific controls: interpretability, agent security, verifiable method.

The research projects

Scalpel: steering an LLM from a single featureInterpretability · AI Safety

Scalpel: steering an LLM from a single feature

Scalpel demonstrates, with numbers and scientific controls, not screenshots, that a single interpretable feature inside a language model causally cont…

0.61 vs 0.00GitHub
COAX: automated red-teaming for AI agentsAI Security · Agent Red-teaming

COAX: automated red-teaming for AI agents

COAX attacks an AI agent the way an attacker would, indirect prompt injection, tool abuse, data exfiltration, adaptive multi-turn attacks, and verifie…

OWASP LLM Top 10GitHub

Need this expertise?

Let's bring research into your product.

Book a free call