Thesis

AI, automation, and the future of fire protection.

Fire protection is a field-service business: people, vehicles, inspections, paperwork, and compliance deadlines. The operational layer is where margin and safety both live, and it still runs on spreadsheets, callouts, and institutional memory. This page is the honest version of where I work and where it is going.

Start a conversation See the work

Current // Testing // Future

Three tenses, kept apart.

CURRENT

What is real now

Daily operational work in fire protection: hours, callouts, scheduling, and the paperwork layer that keeps businesses compliant.

Workflows in fire protection: hours, callouts, scheduling, and the paperwork layer.
Administrative workflow automation for tasks such as hours, scheduling, and reporting; public details remain intentionally general.
Public work connected to this problem space includes browser AI, agent memory, and workflow utilities.

TESTING

What is being tested

The gap between a working workflow and a validated one. Nothing here is claimed as deployed capability.

AI assistance for administrative workflows; no deployment or safety outcome is claimed.
Route optimisation against real constraints rather than demo data.
Local AI infrastructure that runs models close to the work instead of renting every layer.

FUTURE

Where this is headed

The long-term direction. Nothing in this zone is present capability. Robotics is a destination, not a present tense.

Recurring service models where software, inspection data, and maintenance cycles compound.
AI-assisted decision support that leaves certified judgment with people.
Physical automation and robotics where the economics and the safety case justify them.

The argument

In six parts.

01
The industry problem
Fire protection is a field-service business: people, vehicles, inspections, paperwork, compliance deadlines. The work I observe full-time at Auscoast Fire Services runs on spreadsheets, callouts, and institutional memory. The operational layer is where friction accumulates; that is what I pay attention to.
02
Why the workflows are inefficient
The work is repetitive in the places nobody notices: re-keying hours, rebuilding the same report, re-routing the same day. Each step is small; the sum is a second job nobody applied for.
03
Where AI actually helps
AI helps with the mechanical parts: summarising, extracting, routing, drafting, reconciling. It helps with administration and decision support. It does not sign off on compliance, and it should not.
04
What stays human-controlled
Certified inspections, compliance decisions, safety calls, and client relationships stay with people. The system exists to make those people faster and less tired, not to replace their judgment.
05
Recurring service models
Fire protection has recurring inspection and maintenance cycles. A future direction is to combine service, data, and software around that cycle; this is a direction, not a current capability.
06
Evidence and uncertainty
What is certain: the friction is real, the field is real, and the tools work in controlled use. What is uncertain: measured savings, adoption at scale, and how far automation can responsibly go. This page will be updated as evidence replaces intent.

Read this first

AI assists administration, workflow, and decision support. It does not certify. Inspection sign-offs, compliance decisions, and safety calls remain with certified people. Nothing on this site claims a safety outcome that has not been measured, and nothing here implies regulatory approval. Where a claim is uncertain, it says so.

How the work gets built