The AI ops manager for Australian pest control

What the Ops Manager does

• Procedures drafted from your actual job data

  Hunter reads your completed jobs, the checklist steps your crew actually ticks, the callbacks and the corrective actions, and writes the first draft of the procedure — stating those counts as its evidence. You edit it and publish it, and publishing is approval-gated and records a named person. It never publishes itself, and the draft says on its face that nothing in it is a regulatory requirement.

• Approvals, so only the exceptions reach you

  Thresholds you set, on discounts, refunds, credit notes and the actions you would rather see first — with role scoping so an owner is not asked to approve their own work, and the strictest matching rule winning. Everything below the line just happens.

• It handles the routine and briefs you on the rest

  Routines are the substrate the whole product runs on — scheduled or triggered by an event, gated on a signal so they cost nothing when there is nothing to do, and clamped by a trust dial that starts by drafting and earns the right to send. Everything else is a notification you can read later.

• The business on one screen, without touching it

  One screen, one tap from anywhere: what needs a person, what is booked, what has been invoiced and paid, what is at risk. Ask the same question from the ute and you get the same numbers — there is one implementation per figure and both the screen and the agent call it. That is the differentiated claim, and it is literally true: every competitor's dashboard and every competitor's chatbot are separate systems that routinely disagree. The tiles are current as of the moment you opened it; three of them read an overnight snapshot and print the date they were measured.

• A dated plan for getting out of the day-to-day

  Off the tools, then off the phone, then off quoting — three goals with dated milestones. The milestones that matter complete on a measurement rather than on a tick: owner labour minutes falling to zero over a month, owner-answered calls reaching zero, quotes authored by the owner reaching zero. A measured milestone has no done button and cannot be ticked by the agent. Nothing on this screen refers to selling the business or what it might be worth; that is a different surface with its own rules.

• What is actually holding the business back

  Leads and their trend, quote conversion, capacity utilisation, repeat rate and average job value, each tested against a stated threshold, producing one verdict: leads, conversion, capacity, retention or price. The arithmetic is deterministic and shown, so you can disagree with it — the model narrates the answer and may not produce it, re-rank it, or name a constraint the arithmetic did not name. The narrowing: it is a weekly snapshot plus a fresh measurement whenever anybody opens it, not a continuously running analysis, and opening the report never silently manufactures history.

• The run sheet telling you when to hire

  Booked hours over rostered hours, with the backlog and how far out your first free slot is. Two narrowings, both on the screen: where a technician has no working hours entered, the row is measured against an assumed week and says so, and the recommendation only reaches hire after three consecutive weekly snapshots above the threshold — so a busy fortnight cannot become a hiring decision. The reverse direction is deliberately asymmetric: quiet weeks produce 'there is more capacity than work', never a suggestion about anybody's employment.

• Callback root causes, by technician, job type and product

  A callback is linked to its original job, given a root cause, and reported across all three dimensions. This is the single most useful quality number in a pest business and almost nobody records it, because the callback gets done as a favour and never written down.

• Warranty exposure, and a reserve for it

  What cover is in force, what it is ending, and a monthly reserve computed from the claim rate your business has actually seen — recorded against the month and never restated afterwards. Two things it does not do: it does not model what a warranty costs to honour, because the exposure figure is the sum of your own per-warranty estimates, and with no claim history at all the reserve is zero rather than a percentage somebody invented. It moves no money and opens no account.

• The relationships belong to the business

  Calls, texts and email land in one inbox against the contact rather than on a technician's personal mobile — and grouped by the person, so somebody's web enquiry and their reply to your email are one conversation rather than two. When a technician leaves, the customer relationship stays.

• A watch on the rules that affect your products

  You install the searches that matter — the pesticides register, your state's rules, the termite standards, the workplace chemical guidance — and Hunter re-runs them on each watch's own frequency, telling you what looks like it moved with a link to the page it came from. It never edits a compliance rule in your account: a change that needs action raises a task for a human. There is no regulator feed, no completeness claim, and never a summary without its source, because a regulatory claim with no source is a rumour.

  What we do not claim

  The offer originally promised: A live APVMA registry lookup inside the product. The register is a search page, not an interface, and scraping a regulator's registry is brittle and an unchecked question of their terms.