01Overview

Defense-in-depth security for on-device AI.

A security SDK that protects on-device AI agents from prompt injection, unauthorized actions, data leakage, and model theft.

02What we do

A guardrail for on-device AI.

A small SDK that sits between the user, the model, and any action the AI takes. Four runtime checks, all on-device, no network dependency.

InputGuard

Detects prompt injection before it reaches the model.
ToolGuard

Enforces policy on AI-initiated actions: tool calls, transfers, messaging, data sharing.
OutputGuard

Filters sensitive data and detects when the model has been manipulated.
ModelGuard

Protects model weights from extraction and theft.

03Who it's for

i.

Medical device manufacturers

Driven by EU MDR and the EU AI Act.
ii.

Fintech with on-device AI agents

DORA compliance, transaction-capable assistants.
iii.

AI control systems

Manufacturing automation, energy, robotics, autonomous machinery.
iv.

Connected-health platforms

Patient data, on-device triage, remote monitoring.
v.

Automotive in-cabin AI

Voice assistants, driver monitoring, infotainment agents.

04Contact

For pilots, partnerships, technical discussion, or general inquiries.

LocationOslo, Norway

StagePre-seed