Every business already runs on a substrate. Most of them inherited theirs by accident.
The layer underneath — the data nobody owns, the handoffs nobody wrote down, the decisions that live in one person's head — quietly decides what the business above it can and cannot do. Change the layer and the result changes with it. Leave it alone and no amount of effort on top will hold.
So we start underneath. Applied AI systems designed, built and run by operators and engineers, fitted to your customers, your data and the judgement your team already exercises. Not a pilot. Not a demo. The layer the work stands on.
And we go first. If a system cannot survive our payroll, our customers and our margins, it does not get recommended to yours.
- Find the real constraint
- Customer intelligence
- Return time to the team
- Operational flows
- One dependable source
- Structured data
- Prove what improved
- Claims safeguards
- Decision systems
- Substrate Site Builder
Four foundations, laid in the order they carry weight.
Applied AI works when it is fitted to the operating reality: your customers, your data, your team and the decisions they make every day. Strategy names the constraint. Automation returns the time. The data foundation makes it repeatable. Observability proves it moved. Skip one and everything above it sags.
Four layers, dealt in order — strategy, flow, data, measurement
Each one is a real deliverable with an owner, not a workstream on a slide. We lay them in sequence because the one below decides whether the one above can hold any weight at all.
Every system we run reports on itself, out loud
Instrumented at the source, so drift arrives as an event on a Tuesday rather than a surprise in a quarterly. The cards here are a simulated feed built for this page — the real ones sit inside client systems, where they belong.
One surface, the whole operating picture
Quotes, calls, forms, reviews and margin land in the same view, with the source attached to every figure. The point is not the dashboard. The point is that an owner can finally see the shape of their own week.
Lost-quote exposure, priced across a year
Nothing renders until it can be checked
Every claim carries its source and its framing — client work or our own business, never blurred. If a number cannot be reconciled against the live system it does not ship, on our site or on yours.
One diagnostic. One Christchurch trade business. Two years of quotes.
The owner arrived with a feeling rather than a number: the big jobs were slipping and nobody could say why. So we read the ledger — twenty-four months of their live customer quotes, parsed, matched to outcome and reconciled. Everything below is what we found for that client: their trading, their numbers, not ours. They kept the dashboard.
Quotes analysed in one diagnostic
Twenty-four months of the client's live quoting, read end to end rather than sampled.
Client workWin rate across the ledger
The client won 61% of its smallest quotes, under $500. On jobs over $25,000 that rate fell to 36%.
Client workA year in lost-quote exposure, quantified
Priced across a year, the client's lost-quote exposure ran to roughly $1.59 million. The pattern had been sitting in the ledger the whole time.
Client workLive verticals of our own
Yes, these are our own businesses. That is the point — the systems meet real payroll, real customers and real margins before we recommend them elsewhere.
Own businessThe number was always in the ledger. Nobody had ever read the whole thing.
One layer. Five businesses of our own standing on it.
Nutrition coaching, a gym, a clinic, a martial arts academy and a supplements brand. Five verticals running AI for revenue, all drawing on the same foundation. Different customers, different regulators, different margins — and underneath every one of them the same data model, the same safeguards, the same measurement loop. Each surface below shows the same system under a different set of operating conditions. That is the whole argument for building a substrate rather than five clever tools: what you learn in the clinic on a Thursday shows up in the gym on the Friday.
Readouts shown on these surfaces are simulated telemetry, generated in the browser to illustrate the shape of the system. Our own businesses; verified operating figures live on the work page.
Diagnose, design, build, run. Then round again.
We begin with your operating evidence, not a predetermined tool. The four stages stay connected so every decision keeps a reason underneath it.
Engagements? Fair question.
We do not publish rates, and we are not going to pretend a web page can price your problem. Engagements begin with a diagnostic: a bounded piece of work with a fixed scope that leaves you holding a model of your own operation, whether or not you build anything further with us.
If the evidence says the highest-leverage move is something we do not build, we will tell you that and hand you the analysis anyway. It is a short conversation either way, and you keep what comes out of it.
No tiers, no packages, no seat counts. The scope comes out of the evidence, not out of a table.
A note from the three of us.
Substrate exists because we got tired of buying advice from people who never had to live with it. Between us there are twenty-two years of operating businesses, a career of enterprise engineering, and production multi-agent systems that have to work on a Monday morning. We run five companies of our own, and every system on this page met our payroll before it met anyone else's.
That is not a credential, it is a constraint we chose. It means when we tell you a thing is not worth building, we are the ones who already paid to find out. If that sounds like the sort of people you want underneath your business, the diagnostic is where it starts.
our own
businesses
first
Bring the number that doesn't make sense. We'll find the layer underneath it.
You do not need to know the system yet. Bring the evidence, the friction and the consequence — we will help you find the first useful move.
Substrate is an applied AI consultancy in Ōtautahi Christchurch. We engineer the systems New Zealand businesses run on and cultivate what grows on them. See the work, read the method, or just say hello.
Proven in our own companies first, from the bottom of the South Island.
Substrate, the layer things grow on. © 2026
