Software built for one trade, in one country

Pest control is a licensed trade that runs on records, and the software the industry inherited treats those records as an afterthought. A general job-management app will happily let you attach a photo and type some notes. It will not tell you that the record you just saved is missing the two fields your state requires, that the technician who signed it holds a licence that expired last month, or that the copy the customer is legally owed was never sent.

That gap is not a missing feature. It is a modelling problem. Compliance in Australian pest control is per state — different required fields, different deadlines, different retention periods, different rules about who gets a copy and when — and a system with one generic "custom fields" box cannot express it. Every operator using a generic tool ends up building the rules themselves, in spreadsheets and habits and one person’s memory, and then policing them by hand forever.

So Hunter starts from the regulated object rather than bolting it on. A chemical application is a first-class record validated against a state profile. An asset — a bait station, a termite monitoring system — has a service level that drives recurring work. A technician’s licence is a dated thing that expires. Everything else in the product, the quoting and scheduling and invoicing you would expect from any field service tool, is built around that spine rather than the other way around.

Hunter is for Australian pest control businesses, from a sole operator with a ute through to a multi-technician company running residential, commercial and termite work across a state. It is not a general trades app that happens to have a pest module, and it is not an enterprise platform that needs a consultant to configure.

The two people it is designed around pull in opposite directions, which is most of what makes the product what it is. The owner or office manager needs the business to keep moving without them touching every message — quotes chased, jobs booked, invoices sent, payments reconciled. The technician needs the opposite: almost nothing on screen, on a phone, often with one bar of signal, in a roof cavity. Hunter’s field surface caches the day’s work and queues what you do offline; the office surface leans on the agent.

It is a poor fit if you want software you configure once and never speak to, or if your business is not pest control. Hunter’s opinions — treatment cycles, chemical registers, state compliance profiles, the standards this trade works to — are load-bearing, not decoration, and they are the wrong opinions for a plumber.

Most software of this kind grows by adding screens. Every new capability arrives as a new page, a new form, a new tab nobody finds, and the product slowly becomes a filing system its owner has to operate. Hunter is built the other way round: there is one AI agent with tools over every object in the business, and the interface exists for the things a person genuinely does better by hand.

Practically, that means you can ask for what you want — reschedule Thursday, quote the Riverstone job like the last one, tell me who hasn’t paid and chase them — instead of learning where it lives. It also means the agent and the interface run on exactly the same code. Every action has one implementation in a service layer that both the screens and the agent’s tools call, so there is no second, weaker path that the agent takes and no behaviour that exists in the chat but not on the record. If an invoice is created by the agent it is created by the same function that creates it when you press the button, with the same validation, the same numbering and the same entry in the activity log.

The screens that remain are the ones where a person is faster or safer than a sentence: a schedule you drag, a run sheet on a phone, an invoice you scan before it goes. Everything else earns its place or doesn’t get built.

An AI agent that can send messages and move money is only as good as the moment before it does. Hunter’s rule is that the agent may read, draft, plan and prepare freely, but anything that reaches a customer or moves money stops and asks.

When the agent wants to send an email or an SMS, it produces the whole thing for approval — the exact recipient, the exact channel, the exact words — and you approve, edit or reject it. Not a summary of what it intends to send: the message. The same gate sits in front of refunds, charging a saved card, issuing a credit note and importing another system’s ledger into yours, and automated routines carry their own tool policy and dollar limits, so an automation that exceeds what it was trusted with degrades into a proposal rather than proceeding quietly.

This is a deliberate ceiling on autonomy, and it costs the product something. It is also why an operator can turn the thing on in week one. Trust is earned per capability, not granted up front, and the settings that widen it are the operator’s to move.

It is easy to claim a country and mean a currency symbol. Here is what it means in Hunter, item by item — each of these is a decision in the product rather than a localisation setting.