Blog

Ideas on
AI security.

Thinking about browser agent isolation, zero-trust automation, and what it means to build systems where trust doesn't matter.

Technical Deep Dive June 2026 7 min read

What If the Agent Never Had Your Data?

Three layers of architectural security for AI agents that don't require trust. How contextual data isolation, human-in-the-loop control, and immutable browser policy make prompt injection irrelevant.

Read →
Security Model June 2026 5 min read

Your AI Doesn't Need to Know Who You Are

Browser agents can see everything you can see. That's the problem. Contextual Data Isolation changes what the AI can access at the architecture level. The data is removed before the model ever sees it.

Read →
Thought Experiment June 2026 3 min read

The Two Room Thought Experiment

A simple model for reasoning about AI agent security. What happens when you separate the room where work happens from the room where data lives?

Read →

More posts coming. In the meantime, read about the architecture behind RedactSure.

How the security model works →