Available for new work AI Automation Specialist Gold Coast, Australia
I run 40+ AI agents for a dozen Australian small businesses. Everything is measured on real usage and real cost. What your team stops using gets switched off, sometimes the same day you ask.
Status: 40+ agents online across 12 machines, watched daily.
Every figure below comes from systems I built or run in production. Nothing is projected, nothing is rounded up.
Named, watched, inventoried
9 Mac minis in client offices, 3 cloud servers
Each one flagged as used or unused
After diagnosing a silent 19 day outage
Square review, including a double charge path
EasyInspect, live on the Apple App Store
Cost per AI reply
$1.34
Reply latency
16s
Dashboard load
Instant
Before and after, read from token logs and production monitors. Not from a slide.
01The pain
02The control
I sit with your sales, marketing and operations people and find the hours being wasted. Then I delete those hours with scripts, integrations and agents. No tool tours, no platform pitch: the starting point is the job your team hates most.
One client's assistant reads nine business systems, two of them through browser automation because no API exists. The owner asks one question instead of opening nine tabs.
AI is also how you get your business back. Your numbers, your leads and your systems, readable in one place, so decisions stop depending on whoever holds the password. You see what I see.
One owner gets a morning brief at 08:15 with exactly what needs a decision that day. On a 50 property portfolio, no price goes live without a human click.
If it does not remove a real pain or hand you real control, I will not build it.
Usage telemetry on every automation and cost per reply computed from token logs. If your team quietly stops using something, I see it, and I tell you.
I traced a clinic's surprise $717 bill to a default heartbeat burning $165 a day, and shut it down the same week. You will always know what your AI costs to run, per reply, per month.
When an owner asked to stop everything automatic, all 14 scheduled jobs were off the same day. My fleet inventory flags which of 13 deployments earn their keep and which should die.
I killed a daily SEO report that turned out to be hallucinated. I documented my own silent 19 day outage and recovered all 112 leads it lost. You get the honest version, always.
Every deployment ships with a documented stop procedure the owner can run alone. No calling me, no waiting, no hostage systems.
Contract manager on the Airbus aircraft and ATR accounts. Over 220,000 euros of revenue in twelve months at margins above 28 percent, while finishing valedictorian of my business school cohort.
Education agent placing international students: 50+ placements in six months. Built my first AI process design there, and watched non technical advisors actually adopt it. That hooked me.
Tech lead across River House and Crown BNB. Took EasyInspect to the Apple App Store: React, tRPC, MySQL, Stripe, about 40 AI vision calls per inspection, load tested to 10,000 users.
Running a fleet of 40+ named AI agents on 12 always-on machines for a dozen Australian SMEs. Nine Mac minis living in client offices, three servers in the cloud, every one of them watched.
Property inspection platform, live on the Apple App Store. Statutory report formats for all 8 Australian states and territories, about 40 AI vision calls per inspection, Stripe billing, offline mode for inspectors in dead zones.
All 8 Australian states covered
40+ named agents answering email, phone and chat for a dozen Australian SMEs, on machines that live in their offices. Inventoried, measured, and pruned when unused.
Answering 24/7 across a dozen SMEs
Event and function management for a riverside venue: enquiries, quotes and payments in one place.
Replaced a SaaS subscription
Meta Lead Ads to CRM pipeline with a watchdog that speaks up when the flow goes quiet.
112 lost leads recovered
Order and pay takeaway channel on Square for a riverside kiosk, taking real payments from day one.
25/25 security findings fixed
3 SMS and 2 emails after every enquiry. Idempotent, retry safe, and built to fail without double sending.
One command kill switch
Shared memory vault for an owner's AI agents, plus a deterministic morning brief with the day's decisions.
One DM at 08:15 daily
Maintenance job tracking for a property group, from tenant report to resolved and paid.
In production
Most agencies build dependence. I build the opposite: four commitments, in writing, in every engagement.
Code, data, accounts and API keys, in your name from day one. If we part ways tomorrow, everything keeps running and everything stays yours.
ownership: transferred at setup, not at exit
Every system ships with a stop procedure you can run without me. When an owner asked to pause everything, all 14 jobs were off the same day.
kill switch: included, documented, tested
Plain English runbooks, not a folder of code. A non technical intern runs one of my production servers behind five fixed commands.
runbook: written for humans, not developers
Standard tools and boring hosting. Any developer can pick up what I build cold, and the documentation proves it.
stack: open, portable, replaceable
Tell me the job your team hates most. I will tell you honestly whether AI can remove it, what it would cost to run, and what I would refuse to build.
fit check it goes both ways
This form filters, it does not funnel. If I am not the right person for your problem, I will say so and we will not work together. If I am, you get a one page plan with the run cost.
I read every one of these myself. If we are a fit, you will hear from me within a day. If we are not, you will hear that too.
One email about the job that wastes the most hours. Reply within a day.
Thirty minutes about the pain, not the tech. No slides, no jargon.
A one page plan with the run cost, or a straight answer that AI will not fix this.
✦
✦
✦
✦
✦
✦
✦
✦
✦
✦
✦
✦
✦
✦
✦
✦
✦
✦
✦
✦