Three things we build. Most people need the first one.

Automation into your existing stack

Your systems stay. We build what goes between them.

Supplier bills read and posted. Quotes becoming jobs becoming invoices without re-typing. Enquiries routed to the right person. Reminders that go out without anyone remembering to send them.

Cheapest of the three, fastest to value, and what most businesses actually need. It is also the one we recommend when a client arrives asking for a rebuild — the reasoning is on custom vs off-the-shelf.

The reason it is cheapest is not that it is easy. It is that the surface area is small: one connection, defined inputs, defined outputs, and a failure mode you can actually see. A replacement touches everything your team already knows how to do, which is why it reliably costs more than the licence you were trying to escape.

It is also the reversible option. Change job management systems in two years and you rebuild one connector rather than a platform — which matters more than it sounds, because most businesses do change something within five years and almost none of them plan to.

The nine systems we have built against

• XeroAccounting
• MYOBAccounting
• QuickBooks OnlineAccounting
• ServiceM8Job management
• SimproJob management
• StripePayments
• GoHighLevelCRM and marketing
• ZapierAutomation platform
• n8nAutomation platform

Nine is the whole list, not a highlight reel. If your system is not on it we will say so rather than imply experience we do not have — and if it has an API we can still work with it, we just will not pretend we have done it before.

Custom systems, when nothing fits

Worth it under three conditions: the process is how you compete, the workarounds cost more than a build, or nothing exists for what you do.

Most businesses meet none of them. About 80% of our engagements end up custom — which sounds high until you remember that the businesses who reach a build have usually been filtered by a Leak Check that told the others not to.

Seven have been built so far as reference implementations, one per trade — Kindra Sign Shop, Kindra Build, Kindra Electrical, Kindra Plumbing, Kindra Mechanic, Kindra Care Desk and Kindra Clean — each shaped around a specific trade rather than adapted from a generic template.

AI agents, and when they’re actually warranted

Rules-based automation is cheaper, faster and more predictable. Agents earn their cost only where the input is unstructured or the task needs judgement: reading a supplier invoice that arrives differently every month, triaging an inbox, answering a call at 11pm.

If someone’s selling you an agent for work a rule could do, they’re selling you the expensive version.

The distinction is not academic. A rule costs the same to run whether it fires ten times or ten thousand; a model call carries a unit cost, so the economics invert at volume. Anything with a fixed, knowable answer belongs in a rule for that reason alone — and we will move work out of the AI step and into rules wherever we can, which reduces what you pay monthly.

What crosses the boundary when an AI step is involved, and what does not, is set out on ownership and security — including the parts we cannot claim.

What we don’t do

Naming this is a credibility signal, so here it is plainly.

• Website design and brand work
• Marketing campaigns and ad management
• General IT support and helpdesk
• Anything requiring a supplier assessed against the Australian Privacy Principles — we have not been
• Projects needing a team of twelve and a project office

We will point you elsewhere for these rather than take the work and learn on your money. The filter runs on the first call, which is the cheapest place for both of us to find out.

Which one you probably need

• Systems are decent but don’t talkAutomation. This is most businesses, and it is the cheapest of the three.
• Your real process lives in a spreadsheet beside the softwareAutomation first, custom later — and often the automation turns out to be enough.
• Your process is genuinely unlike anything a product supportsCustom. Rare, and worth testing against the framework before committing.
• The bottleneck is a human reading unstructured thingsAgents — narrowly, for that step, not for the whole workflow.

The Leak Check sorts this in twenty minutes, and how an engagement runs sets out what happens after it.