Agents become the interface.
The next layer of the internet is mediated by software that acts on our behalf. The infrastructure that makes those agents auditable, governable, and trustworthy is undervalued and under-built.
Applying two decades of experience to frontier AI
DANZUS
Builds quietly.
D · AU · NZ · US
AI-native software for the trades, compliance, and the operators who run the physical world.
A privately held software studio. Six active projects across AI, compliance, and the trades.
§ 002 / Thesis
Three beliefs.
Intelligence moves to the device.
Inference belongs where the work happens. The applications that matter in the next decade are designed for real environments — workshops, job sites, classrooms, the cabs of trucks.
Operators need better software.
Robots are the future of the factory. Humans are the future of the infrastructure. The next generation of welders, drivers, electricians and inspectors deserves software built for the way the work actually happens.
§ 003 / Work
What we've shipped.
Active ventures — six.
-
/01
DimeVision (incl. BeadBuilder)
An AI mentor for welders. Computer vision scores beads on a phone, Danny Dime coaches in plain English, and BeadBuilder simulates GMAW before students ever strike an arc.
-
/02
Agent Audit Trail
A tamper-evident logging layer for AI agents. Hash-chained records and a policy engine, delivered as an MCP server. Built for SMBs facing Colorado SB 205 and the EU AI Act.
-
/03
OSHA Compliance MCP
A Model Context Protocol server giving AI agents structured access to OSHA standards. Open infrastructure for compliance-aware automation in the trades.
-
/04
DOT / FMCSA MCP
The same playbook applied to commercial transport. Federal motor carrier rules, hours-of-service logic, and inspection criteria — exposed to agents that need them in the cab and the back office.
-
/05
Ed (clawdbot)
An in-house operator agent running on a Mac Mini M4. Handles routine engineering work across the portfolio under a tiered architecture that reserves higher-judgment tasks for Claude.
-
/06
ContextKit
A free interactive guide to the chatbot-to-agent paradigm shift. Write core files and skills that turn any AI — ChatGPT, Claude, Gemini, Cursor, Claude Code, or OpenClaw — into an agent that does the work, not just answers questions.
§ 004 / Operator
Who's behind this.
Five industries across three countries. Aircraft maintenance in the Royal Australian Air Force. Subsea quality assurance on Chevron's Wheatstone project in Western Australia. Eight years building analytics and workforce strategy for Fortune 500 companies in the US. Data visualization. And now, AI-native software.
The throughline is standards, repeatability, and close attention to what actually works — applied creatively to whatever frontier technology makes the work better.
DANZUS is the holding structure. The work is experimental — applying hard-won experience from hangars, job sites, and enterprise data to problems most AI companies aren't looking at.
§ 005 / Signal