Not currently accepting new engagements

I make AI prototypes survive contact with real users.

Security, observability, the boring stuff that decides whether it ships. Most organisations are sitting on prototypes that will never go live. I close that gap.

Hi, I’m Aggy20 yrs at the sharp endOn-site when it mattersSignal only

The map · four stages of AI in practice

01
Chat
02
Workspace
03
Agentic
04
Production

Most stop at one. The last is where I work.   Open the map ↗


I learn, make, break, fix. Two decades at the sharp end of IT — every hat, from sysadmin to strategist.

I’ve built and defended digital infrastructure for global financial institutions, government agencies, and major corporations. Environments where failure isn’t an option.

I’ve watched every major technology wave arrive with the same pattern. Extraordinary in a demo. Fragile in production.

“If you have a hair-on-fire problem, I provide the path out of the flames.”

Agentic AI — systems that read your codebase, propose a plan, execute it without you in the loop — is no different. The demos are extraordinary. The production deployments rarely happen.

I close that gap. Not with a framework or a methodology deck. With the discipline of hardening systems that cannot afford to fail: correctness, observability, security, maintainability.

If your AI initiative needs to move from interesting to indispensable, you know where to find me.

Direct line: Message on Signal ↗

Signal only. Encrypted, open source, no metadata.
The same channel I use for everything.

Recent thinking

From the journal

All entries ↗
Field work

Projects