I build security, design it, and get the call when it breaks.
This is where I write up the work: the hardware and agents I run myself, the frameworks I use to judge AI security, and what incidents taught me. Every post is one full write-up. Examples are fictional and sources are credited.
- Project
- Field notes on security
- Drawn by
- Blue Marshal
- Disciplines
- Engineering, architecture, response
- Posts
- 3
- Latest issue
- November 12, 2026
Latest posts
- Post 3
AI visibility: which tool can actually see your AI?
An AI inventory that only lists what you approved misses most of what's running. The AI Visibility Map names the tool that sees each kind of AI in each part of the estate, and turns the gaps into an action list. With what Microsoft Defender, Palo Alto, CrowdStrike, Tanium and others can discover today, a worked example and a free template.
- Post 2
AI security maturity: find your stage, then take the next step
The SANS AI Security Maturity Model gives you five stages across three pillars and a scoring method that stops you from overstating where you are. Here it is mapped to MITRE ATLAS and the NIST AI RMF, with a worked score and the next step from each stage.
- Post 1
A 125B model on a DDR4 desktop
Two Intel Arc Pro B60 cards on a budget DDR4 board run a 125B model at about 40 tokens a second. The parts, the PCIe layout, the software stack from the driver up, and what I still send to Claude.