AIAF is the agentic environment I build and run — a working stack of AI agents and automations that handle research, content, design, marketing, operations, and deployments across multiple ventures every day. Not a demo. A system already in production, predicated on one rule: automate tasks that AI handles well, and keep the human in the loop exactly where judgment, relationships, and accountability live.
AIAF is run by one person and a stack of AI systems, working as a command structure — and the transparency is the point.
The judgment and the accountability. Every consequential call — and every "keep this human" — is mine.
The orchestration layer. Runs operations against a written operating constitution — routing work, holding the standards, keeping the whole stack pointed at the outcome.
Claude, GPT, Moonshot, and Gemini run together as a standing council — they research, draft, and cross-examine each other so no single model's blind spot goes unchecked.
The method: a human commander, an AI COO, and a multi-model team that argues before it acts. The strategy behind this very page was pressure-tested by the council before it shipped.
AIAF is a body of working agentic infrastructure. The pieces that make it run:
Every automation starts by breaking a process down task by task and sorting each one honestly: automate it, augment it with AI, or keep it human. The analysis decides; the build follows. Nothing gets automated just to have AI in it.
Automations ingest expert AI and technology sources on a schedule, distill each new item into a structured entry — the claim, its evidence quality, whether it's worth acting on — and archive it to a version-controlled store. Every few hours, automatically.
Research, content, design, marketing, operations, and deployment workflows running on schedule and on trigger across multiple ventures — with human approval gates in front of anything that spends money, sends a message, or goes public.
A permission system and honesty guardrails that let the agents operate on their own without operating recklessly — the difference between a workforce you trust and a demo you babysit.
The principle that shapes all of it.
Map the real workflow — the one on the floor, not the org chart — and decompose it into logical tasks. Each gets an honest verdict: automate, augment, or keep human. Where judgment carries the value, the human stays by design.
Only what the analysis justifies gets built, documented as it's built, with approval gates in front of anything consequential.
The goal of any system is that it runs without babysitting — engineered for the outcome it was named for, not for how much AI it contains.
Operator / Automation Engineer. Almost forty years of hands-on operational experience — analyzing processes, fixing them, automating them — first in industrial and business automation, now in AI-driven agents and workflows. I've lived through enough operational instances to know that the “way it's always been done” does not always equate to the best way forward.
That's what AIAF is built on: workflow breakdown and analysis. Logically deconstructing a process into its singular tasks and reading each one correctly — this piece is redundant or mechanical and should be automated out of the humans' hands; this piece is judgment and should never leave a human's guidance; it's the discipline that decides whether automation helps you or just complicates the task. No tool shortcuts it.
I run my own ventures on the same agentic infrastructure AIAF is made of: research, content, design, marketing, operations, and deployment pipelines in daily production across multiple businesses. When I say an approach works, it's because it's already survived contact with production. Mine.
If you're thinking about what AI can actually do in a real operation — the engineered version, not the hype version — that's the conversation I enjoy most. One email, a sentence or two about what you're working on. You'll get a real reply, from me.