For business owners

Build the AI personal assistant your business runs on.

Most AI lives in a tab somebody has to remember to open. What we install joins up the software you already pay for, so the repetitive work still gets done on the days when nobody has a spare minute. We build it with you, so you can run it when we're not around.

Built alongside your team
The systems end up yours
Scoped after we map the work

Mappedwe work out which jobs are worth handing to a machine
Installedwired into how the business already runs
Ownedyou understand it, so it keeps working after someone leaves

Why it matters

It isn't a tool. It's the layer underneath the tools.

A quote request lands in the inbox.

Someone reads it when they get a minute. They go hunting for the last similar job to remember what was charged. They write the reply from scratch. It goes out the next morning, or it doesn't go out. No single step is wrong and the whole thing still takes two days.

The same request, on a system.

It gets read, matched against what you've quoted for work like it before, and a draft is sitting there with the numbers already filled in. A person approves it or fixes it. It's out the same hour, and nobody had to remember to do it.

What gets built

Four layers, installed in this order.

The order matters. An agent layer on top of no knowledge layer is a demo, not a system.

Where every build starts

Everything your business knows, in one place AI can reach.

Your pricing logic. Your best past proposals. Your policies. How you handle the customer who wants a refund on day 40. Most of that lives in one person's head, an unsearchable shared drive, or a Slack thread from two years ago. Making it retrievable is the unglamorous first build, and almost every other system depends on it existing.

Usually the first hour/day back

The path from stranger to customer, wired end to end.

The before-and-after further up this page is this build. It is the most common place to start, for two reasons: it is usually where the missing hour a day is hiding, and everyone in the business can feel the difference the week it goes live. Nothing sends without a person approving it — that is a design decision, not a limitation we are working around.

Runs without you

Numbers that show up without anyone assembling them.

The monthly report nobody has time to build: pulled, written, and in your inbox on the first. Built with the same attribution discipline we hold ourselves to — every source separated, so you can tell which channel produced customers instead of guessing. If you can't see where the work came from, you can't decide where to put the next dollar.

Built last, on purpose

Where it stops being automation and starts acting like staff.

Once the knowledge layer and the spine exist, agents can take on work that needs judgment across several steps — triage, research, drafting, chasing the thing that went quiet. This comes last deliberately. An agent with nothing underneath it is a very impressive way to produce confident nonsense.

Why us

A few things we do differently.

You're in the room for the build

Done with you, not for you. You end up owning the systems and understanding how they work — that is the point, and it is also the only way they survive contact with your team.

You own what gets built

The systems are yours, sitting in your accounts, on your data. Not rented back to you, and not switched off the day you stop paying us.

Nothing sends without approval

Every system that talks to a customer puts a draft in front of a person first. That doesn't change once we hand it over.

You deal with me, not a rep

The person who runs the mapping is the person who builds it. It never gets passed to a delivery team you've never met.

How it works

The practical pathway to AI inside your business.

1
Learn where AI creates value, and where it does not.
Understand the technology behind the opportunity
A working understanding of what automations, workflows, models, and agents each really are, and which one a given problem calls for. Not a course — the vocabulary and the decision frameworks, so you can brief your team, evaluate a vendor, and tell the difference between a tool that will change your margin and one that will just be a subscription.
- A plain-language map of automation vs. workflow vs. agent, and when each is the right answer
- Decision frameworks for evaluating any AI tool or vendor pitch
- The language to brief your team and your partners without hand-waving
- A written list of what AI cannot currently do well in your business
2
Connect AI into how the business sells, delivers, and manages work.
Build the layer your company runs on
This is the core, and the least glamorous part of the job. Most companies already pay for software that could do half of it. What they do not have is anything joining it up, so the same customer detail gets typed in three times by three people and the reporting waits on whoever has a free afternoon. The goal is not more software. It is one layer beneath the software you already own — and a team that trusts it enough to stop keeping a private spreadsheet on the side.
- Internal knowledge and documentation turned into surfaces AI can query
- Admin, support, CRM, and reporting workflows built and running
- Sales and delivery connected end to end — no re-typing between systems
- An operating layer your team understands well enough to trust
3
Build systems that fit your process instead of fighting it.
Shaped around how you already work
Off-the-shelf AI apps are built for the average of ten thousand businesses, which means they're built for none of them. The capability we install is the ability to make the specific thing your process needs: an automation, an internal tool, an agent that knows your rules. Including how to scope one before you build it, which is the step most people skip and then regret.
- Custom automations, internal tools, and agents fitted to your workflow
- A scoping method that catches the bad build before you pay for it
- Patterns for turning repeatable work into a system that runs itself
- The judgment to know when off-the-shelf is genuinely the better buy
4
Find the highest-value opportunities before spending on any of them.
Know what to build first
We sit down with how the work really moves and look for the expensive parts: work that repeats, work that waits on one person, work that gets done late because nobody has time. Each candidate gets sized on impact, build difficulty, and whether your team will realistically adopt it. What comes out is a short ordered queue, not twelve simultaneous experiments.
- A written audit of where the hours go and which person everything waits on
- ROI estimates tied to hours, margin, output, and response time
- Prioritisation by impact, build complexity, and honest team readiness
- A focused build queue — first, second, third, and explicitly not yet
5
Turn the capability you just built into something you can sell.
Only if you sell to other businesses
Optional, and it does not apply to everyone. But if you sell services to other businesses — agency, consultancy, MSP, bookkeeping — you finish this holding something most people pitching AI do not have: a working system you built, in a company you understand completely. That is a much easier sale than a deck about somebody else's case study.
- An AI audit you can offer to existing clients, packaged and priced
- Implementation projects scoped around outcomes rather than hours
- Ongoing support structures that produce recurring revenue
- Positioning for clients who already trust you and don't need convincing

Fit check

Is this right for you?

This is for you if…

You want AI installed in the business, not another tab open.

You're not short of prompts, apps, or things to try. What you want is the work getting done on a Tuesday when everyone is busy and nobody opens a chat window.

There's manual work you can already name.

Admin, follow-up, reporting, CRM updates, proposals, quoting, support replies, documentation. You can see the work your team repeats every single week, and you want it gone rather than delegated.

You need someone to tell you where to start.

You're fairly sure AI could help and completely unsure where to start. What you need is a ranked list — biggest win first, honest about what each one costs to build — not another tool trial.

You want to end up owning it.

You'd rather understand the systems your business depends on than rent them from someone who disappears. Everything built here is yours, and you'll know how it works.

You want it measured.

Hours saved. Response time down. Margin up. Tasks that no longer touch a human. You want to look at a number afterwards, not at a demo.

You're willing to look at how the business actually runs.

The mapping session goes into your real processes, including the embarrassing ones. That is where the expensive habits hide, and it is the part that cannot be skipped.

This is not for you if…

You want it done for you, with no involvement.

This is done with you. You get the roadmap, the builds, and the support, but you still have to make decisions about your own business and help install things properly. Nobody can do that part from outside.

You're shopping for the newest tool.

If you're after the current month's most exciting app, this isn't it. The work here is closer to plumbing: less fun to talk about, and it lasts a great deal longer.

You want AI to replace your team.

That's not the work we do. These systems take the chasing and the re-typing off your people, not your people off the payroll. If a smaller headcount is the goal, we're the wrong fit.

You want a guarantee about the outcome.

We don't give those, here or anywhere else on this site. We'll tell you what we're building, why we think it is the first thing worth building, and how we will measure it. We won't promise you a number.

It's going to stay a side project.

AI does nothing sitting in a tab next to the business. This is for owners ready to change how the team works, not to run an experiment in a spare afternoon.

Nothing about how you work can change.

Some of what we find will be a process problem that AI would only make faster and wronger. If no process can be touched, there's very little here worth installing.

What it costs

One honest number, then it depends.

$2,500

to map your business and install the first system

After that it depends on what the mapping turns up. Nothing gets built before you have seen the scope and the price for it. And if what you need turns out to be smaller than this, the call is where we'd find that out.

Who you'd be talking to

You get me, not a sales team.

I do the mapping and I do the building, and I'm the one sitting with your team while they learn to run it. When something needs changing six months later, you're talking to the person who built it rather than to whoever picks up.

Before you book

The questions people actually ask.

What is an AI operating system?

It's the layer that sits underneath the tools you already use and connects them with judgment. Concretely: one place where everything your business knows is retrievable, a set of workflows that move a customer from first contact to paid without anyone re-typing anything, reporting that assembles itself, and — last — agents that handle the multi-step work needing judgment. The word 'system' is doing the work in that phrase. A dozen clever prompts is not one.

How is this different from just using ChatGPT?

A chat window is a tool you have to remember to open, that forgets your business between sessions, and that produces output someone then has to move somewhere by hand. An operating system knows your pricing, your history, and your policies; it runs on a trigger rather than on someone's memory; and its output lands where the work actually happens. Most businesses have the first thing and think they've adopted AI.

Do I need a technical team or a developer?

No. You need to understand your own business and be willing to make decisions about it. The building is our side of the table. Where something genuinely needs code, we write it — but a surprising amount of the highest-value work is configuration and clear thinking about the process, not engineering.

Will this work for my industry?

The specifics change completely; the shape almost never does. Every business has knowledge trapped in someone's head, a path from enquiry to paid with manual gaps in it, and reporting that happens late because a person has to build it. What differs is which of those is costing you most, which is exactly what the mapping session is for.

Is my data safe? Do I have to hand everything over to AI?

No, and you shouldn't. Part of the design is deciding what a model is allowed to see, what stays internal, and what never leaves your systems at all. Plenty of the highest-value builds touch no sensitive data whatsoever. We'll make those boundaries explicit in writing before anything is built, not after.

Will this replace my staff?

Not how we build it. The work these systems take on is the work people complain about — chasing, re-typing, assembling, remembering. What's left is the part that needs a human, which is usually the part your customers are actually paying for. If your goal is headcount reduction, say so early, because we'd be a poor fit.

How fast will I see a return, and how is it measured?

The first system usually goes in within weeks and is aimed at something you can feel — hours back, or a response time that drops sharply. We agree on how it's measured before it's built: hours, response time, output, margin, tasks no longer touching a human. What we won't do is promise you a figure up front. Nobody honest can.

Do I have to buy a stack of new software?

Usually the opposite. Most businesses we look at are already paying for more tools than they use, several of which overlap. A common early finding is that two subscriptions can go and the remaining one needs wiring properly. Where a new piece is genuinely needed we'll say so and say why.

What does it cost?

$2,500 to map your business and install the first system. After that you get a scope and a fixed price for each build before it starts, taken in the order the mapping put them in. You are free to stop after the first one, and some people should.

In their words

Two owners who ended up running it themselves.

Both of these are chiropractic practices in Australia, and what we built with them were websites — online booking and chat bots wired in, not the company-wide system this page is about. What's worth hearing is how it got built: they both came out of it able to run the thing without me.

“It’s been a game changer for us.”
“We’ve been working with Oz trying to redevelop our websites. It’s so simple — making beautiful websites really quickly, integrating with bots and online booking systems. Oz has been really great and quick on feedback. Really appreciate it.”

Dr Paul CaterPalm Beach Chiropractic, Australia

“Truly remarkable.”
“He’s helped us build our website — we went from being novices to building a full website, integrating bots, getting daily updates. I wouldn’t have been able to do it without his help. If you’re looking to build websites or SEO, he’s been absolutely amazing, and I highly recommend him.”

Dr James ShipwayBanora Chiropractic, Australia

Get started

Start with how the business actually runs.

Tell me how work moves through your business — where it stalls, what gets typed in twice, what only one person knows how to do. I'll tell you which parts of that a machine should be handling and which parts it shouldn't, including when the honest answer is not yet.

30 minutes
No pitch
Just see if we're a fit