Your next constraint is already leaving a trace.
Bring the numbers, the friction and the ambition. We will help you find the layer worth building next, and the first conversation costs you an hour.
Substrate is the layer a business runs on: the data it trusts, the work it repeats, the calls it makes at six in the morning. We engineer that layer for owner-operated New Zealand businesses, and then we run it with you.
Most businesses do not have an AI problem. They have a foundation problem that AI makes impossible to ignore: two systems that disagree about the same customer, work that quietly gets done twice, decisions made from memory because the evidence lives in someone's inbox.
We start underneath. Find where value is actually leaking, build the data and workflow layer that closes it, then instrument the whole thing so it keeps proving its worth after we have gone quiet.
We do this for owner-operated businesses in Aotearoa. We did it in our own companies first, where being wrong costs us rather than you.
The four layers below are not a service menu. They are the order things have to be built in, and we will tell you which one you actually need before anyone talks about scope.
Strategy first, because the wrong constraint is an expensive thing to automate. Then the flow of work, then the data sitting under it, then the instruments that keep both honest. Skip one and the layer above it drifts within a quarter.
Every system we build reports on itself. Events land as they happen, thresholds get named up front, and drift arrives as a notification rather than as a quarterly surprise. The cards here are a simulation of that feed, not a live customer system.
Nobody on the floor wants a model. They want one screen that says what changed, what it cost and what to do before lunch. That screen is the deliverable, and the numbers on it stay attached to where they came from.
We read twenty-four months of a client's quote ledger and priced the pattern underneath their win rate. It had been sitting in their own data the whole time, unnamed and unpriced. They keep the model and the dashboard either way.
Two kinds of evidence appear on this page and we keep them apart on purpose. Client findings are what we found for a client, in their data. Own-business figures come from companies we operate, where the consequence of being wrong lands on our payroll.
If a figure on this page cannot be traced back to its source, it does not ship.
A nutrition coaching business, a gym, a clinic inside a regulated vertical, and a trade business's quote ledger. Different data, different thresholds, different definitions of a bad Tuesday. What does not change is the layer underneath: one source of truth, an instrumented workflow sitting on top of it, and a surface that tells an operator what moved and why it matters. The four panels below are the same diagnostic surface running four different contexts, and every reading on them is invented for the demonstration.
Four stages, connected end to end. Nobody hands you to a different team halfway through, and the people who reasoned about the constraint stay accountable for the build.
We do not publish rates. A diagnostic for a five-person trade business and a data foundation for a multi-site operator are not the same piece of work, and pretending otherwise on a website helps nobody standing on either end of it.
What you will have before any money moves: what we think the constraint is, what we would build first, and exactly what you keep if you stop after the diagnostic.
The diagnostic model, the analysis and the documentation are yours either way.
A note, while you are here.
We started Substrate because we kept building this layer for our own businesses, then watched other owners pay good money for the version that does not work. A chatbot bolted to the side of a website. A pilot that never left the pilot. A dashboard nobody opens twice.
The awkward part is that the interesting work happens underneath, where nobody can see it and nobody can photograph it. A gym's structured-data model is not a good story at a barbecue. It is, however, the reason a machine can answer a question about that gym correctly, and it turns out that matters rather a lot now.
So bring us the number that does not make sense. We will go and look underneath it.
With much appreciation, from Ōtautahi Christchurch.
Bring the numbers, the friction and the ambition. We will help you find the layer worth building next, and the first conversation costs you an hour.