Leads that arrive already scored.
A scored intake gives the team useful context before the first coaching conversation.
Strategy · Automation · Data · Observability
What thrives above is engineered below.
Our belief
Every business already runs on a substrate — the data it trusts, the decisions it repeats, the handoffs nobody questions. We engineer that layer, in companies we own, first.
What we build
The layer underneath, and what it changed.
View all workEngineered to be the answer.
One source of truth fans into structured data, suburb pages and a durable library of useful articles.
Owning the answer inside the rules.
Schema builders, comparison content and a voice linter are run against the site before release.
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.
The practice
Behind every system, a business that has to live with it.
Substrate is an applied AI consultancy in Christchurch. We engineer the systems NZ businesses run on and cultivate what grows on them. Proven in our own companies first.
Operators who understand the business. Engineers who can build the answer. Senior people stay close — from the first diagnostic question to the system running in production.
One diagnostic
One diagnostic. One Christchurch trade business.
Live customer quotes were analysed to expose the commercial pattern beneath the quoting result.
Simulated telemetry — diagnostic replay
Pass 01 · 00:00
The feed above is simulated. The three figures above it are not.
In their words
In their words
01 / 03
Start beneath the surface
Your next growth constraint is already leaving a trace.
Bring the numbers, the friction and the ambition. We'll help you find the layer worth building next.
One reply from a senior engineer, not a sequence. Nothing is sent from this prototype.
Frequently asked questions
With a diagnostic, across weeks 1 to 4. We interrogate the operating evidence behind the problem: variance, margin distribution, conversion economics, customer behaviour and the handoffs around them. No tool is chosen before the constraint is named.
A diagnostic model you keep, with a clear commercial constraint and the evidence needed to act. If the work continues, you also keep the blueprint, the production system and its measurable operating signals.
Each one names the claim, the source and whose result it is. Client findings are labelled as client work; our own operating figures are labelled as our own businesses. Review counts are shown as floors — 1,000+ and 260+ — never as exact totals, and a figure that cannot be verified against its live source does not ship.
Senior engineers, and the same people who reasoned about the constraint. The person who found the problem stays accountable for the implementation, so nothing is handed to a delivery team that was not in the room.
We monitor the operating result, recalibrate the model and improve the system as the business changes. Every system is instrumented to learn from live use, expose drift and compound what works.