A converted warehouse office early in the morning. A man in a slate blue jumper sits sideways at a desk with his chin on his hand, a woman with silver hair stands further back holding a mug, a woman in a terracotta blouse writes on a pad resting on a filing cabinet, and a man in a grey t-shirt sits with his back to the room. The middle of the room is empty pale brick and one tall window.

What process is costing you the most?

We find where AI can create measurable value, implement it inside your existing workflow, and help your team actually use it.

Let's turn that into revenue Not ready to talk? Get the guide

Power your team with the right tools, in the right places, with rules that hold.Most of what you need is already on the market. The work is choosing what goes where, and making it safe for your data, your clients' information and the people using it.

Speed

The turnaround on the work your customers actually wait for. Quotes, reports, correspondence, onboarding, month end.

Cost

The hours going into work that doesn't need a person doing all of it, and the software you're paying for and not using.

Accuracy

The rework, the missed detail, the thing that had to be sent twice. Consistency across everyone who does the job, not only your best operator.

Safety

Your data and your clients' information stay where they should. Your team knows in writing what may go into which tool and what may not, before anything is switched on.

The guide

Maximise your ROI.

Five areas AI can pay off in your business now: quotes, reports, correspondence, onboarding and month end. Then the order we put AI into an Australian business, written down. For an owner or an ops manager who wants to know where it pays before they spend.

The cover of the Anum guide, titled Putting AI into an Australian business, properly and safely, in the right order, with five places AI pays first and then nine chapters listed below it.
What is inside
  • Five areas AI can pay off now: quotes, reports, correspondence, onboarding and month end
  • What actually goes wrong, and the four failures behind it
  • Five questions to answer before you buy anything
  • What can and cannot go into a public model, with the regulator's own position
  • A one page acceptable use policy you can adapt tonight
  • A scoring method for choosing the first use case
  • Training, ownership, and what finished actually means
  • Measuring whether it paid, and when to stop

14 pages, A4. Every legal, regulatory and pricing claim carries the source it came from.

The guide downloads straight away. Your email tells us who is reading it. No sequence, no CRM, no mailing list, and nobody calls you unless you ask.
Anum Pty Ltd collects your name and email for that purpose and no other. Both are held by our website host in the United States. How to get a copy or have it deleted is in our privacy policy.

Here it is

The download should have started on its own. If it did not, the link below is the whole document.

Download the PDF

When you want the version of this built from your own work, that is Discovery. Scope and price are agreed in writing before it starts. Book a 30 minute call.

The problem

You already bought the tools. Nothing about the work changed.

Copilot came with the licence bundle. Someone expensed a ChatGPT subscription. A few people use it for the odd email. The quotes still take as long as they took last year.

01

Nobody owns it

There is no one whose job it is to find the use cases, put them in, and make them stick. It sits with whoever is enthusiastic, on top of their real work, until they get busy.

02

There are no rules

No written policy means the cautious half of your staff avoid it entirely and the confident half paste client information into a chat window. Both of those are a problem, and only one of them is obvious.

03

The training didn't take

A session happened. Everyone nodded. The first busy week, they went back to the old way. Putting AI in is the easy half. Getting people still using it a month later is the work.

04

There is no number

Nobody measured what the work cost before, so nobody can say whether any of it is paying for itself. The subscription renews on the assumption that it probably is.

A large organisation hires someone to own all of this. At the size of most Australian businesses the salary is hard to justify, so it goes undone. That is the gap Anum fills.

How it works

The method, published.

Eight steps, in this order, every time. Most of this industry keeps its method behind a booking form. Ours is on the page so you can judge it before you pay for it.

Software gets installed. Work gets changed by people.

Whether a rollout worked comes down to whether the people doing the job are still doing it the new way a month later.

  1. 01Discover

    We watch the work and write down where time goes.

  2. 02Prioritise

    By value and effort. Anything AI shouldn't touch comes off.

  3. 03Baseline

    What those workflows cost you today, before anything changes.

  4. 04Govern

    Accounts, access, and a one page policy people will read.

  5. 05Implement

    Built or configured inside the workflow people already use.

  6. 06Pilot

    One team runs it on live work first.

  7. 07Train

    We work beside your staff on their own jobs.

  8. 08Measure and roll out

    Against the baseline, then out to the rest.

Two colleagues at one desk, the older one at the keyboard while a colleague points at something on the screen.
Pricing

Discovery first. Everything after it is scoped from what we find.

The first step is a free 30 minute call. We don't publish a price, because the honest one depends on the size of your business and how many teams are in scope. Discovery is quoted in writing before it starts, and the Rollout price comes out of Discovery, also in writing, before you commit to anything.

Two people at a wall of blank sticky notes and butcher's paper, mapping out a process by hand with their sleeves rolled up.
Start here · Phase one

Discovery & Training

The first phase. You come out of it knowing where AI is worth investing in your business, and where it isn't.

  • Every team watched at their real work, and the workflows mapped.
  • What the current way costs you, in your own numbers.
  • The use cases ranked by payback, and the honest answer if there is not one.
  • Your team trained on the AI you already pay for, and on what can safely go near client information.
  • A Rollout proposal, scoped and priced, in writing.

If Discovery leads to a Rollout and you proceed within 60 days, we credit the Discovery fee towards the implementation. If you don't proceed, everything produced through Discovery is still yours.

The eight things Discovery hands you are listed with the contact form.

The Discovery guarantee

If we don't identify at least one AI use case with a clear payback inside six months, we'll tell you the Rollout isn't worth it yet, and refund the Discovery fee.

SCOPED IN DISCOVERY

The Rollout

Where what we found in Discovery gets put into the real work and measured. Finished when your people are still working the new way a month after we leave.

  • Governance first: accounts, access, and an AI use policy signed off.
  • The highest ranked use cases built into the systems your people already open.
  • One team piloting on live work before anyone else is asked to change.
  • Everyone else trained at their own desk, on their own job, until they can do it without us.
  • The result measured against Discovery's baseline, with the documentation, the policy and the prompt library left in your accounts.

The people who ran your Discovery run your Rollout. Scope and price are set at the end of Discovery.

The Rollout guarantee

If fewer than 70% of trained staff are still using the new workflow 30 days after we finish, we come back and retrain at no charge until they are.

This holds where you release your staff for the training sessions we schedule.
SCOPED IN DISCOVERY

Fractional AI Lead

Optional. For what comes after the Rollout: new people, new use cases, new tools. The Rollout stands on its own. This is the next round of work, not upkeep of the last one.

  • New staff trained as they join.
  • The next use cases found and built.
  • New tools assessed as they land.
  • The policy kept current.

Scoped around how much ongoing support you actually need, from monthly advisory through to regular days each week. Minimum three months, then month to month with 30 days notice either way.

We don't sell licences and we take no vendor commission, so a tool recommendation is only ever that.

The Fractional guarantee

Every quarter you get a written read on whether the AI spend is still earning, and that includes what you pay us. If we can't show it is, you can end the engagement there, without the 30 days notice and with nothing further to pay.

What you may already be paying for

Microsoft 365 Copilot Business on 25 seats is AU$8,073 to AU$11,304 a year.

Say half those seats belong to people who opened it twice and never went back. That is about AU$4,000 to AU$5,650 a year buying nothing, and the same cost often runs three times over: a Microsoft bundle, a Claude seat, a ChatGPT seat.

No licence seller measures your split. Discovery measures it.

What the seats cost, and where those prices come from
  • Microsoft 365 Copilot Business AU$37.68 per user per month billed monthly, or AU$26.91 paid yearly. It is an add-on, so a qualifying Microsoft 365 licence sits underneath it.
  • Microsoft 365 Copilot, the add-on for enterprise plans, AU$44.90 per user per month paid yearly, AU$47.15 billed monthly.
  • Claude Team US$20 per seat per month billed annually, US$25 billed monthly, for teams of 2 to 150.
  • Claude Enterprise US$20 per seat per month plus usage at API rates, billed annually. The seat fee covers access and no usage, and the published minimum is 20 seats self-serve or 50 through sales.
  • ChatGPT Business US$20 per user per month billed annually, or US$25 billed monthly, minimum two seats. OpenAI lists and charges these in US dollars. ChatGPT Enterprise is priced on application.

Third party pricing, not ours, from each vendor's own page, checked 24 August 2026. Microsoft Australia and its enterprise page: all figures exclude GST, and the AU$26.91 yearly price is promotional until 30 September 2026, after which Microsoft lists AU$31.40. Anthropic: US dollars, as Anthropic publishes them, with the Enterprise seat minimums from Anthropic's own support page. OpenAI: US dollars, as OpenAI publishes and charges them, with Enterprise the only tier OpenAI does not publish.

What this replaces

A salaried AI Enablement Lead in Melbourne is advertised at $166,939 to $191,084 a year.

That is a live Victorian public sector role, advertised this month. It is the work described on this page: finding the use cases, writing the policy, training the staff, and measuring whether any of it landed.

Anum is that function for the part of the week you need it. It starts the week you engage us, not three months after you start recruiting.

What sits on top of that salary, and where the range comes from
  • recruitment and onboarding
  • leave and equipment
  • the months it takes a new hire to learn your business
  • the risk that the person you hired is the wrong one

Large organisations create that role because the work is worth doing. A smaller business rarely needs it five days a week, and almost never gets to justify the hire.

Advertised range taken from AI Enablement & Capability Lead, Victorian Legal Services Board and Commissioner, August 2026. It matches the published Victorian Public Service grade 6.2 band effective 1 May 2026.

The front counter of a small Australian business, someone behind it finishing a phone call in an ordinary lived-in reception area.
Contact · A 30 minute call

Let's turn your most expensive process into revenue.

Send the form and we write back to arrange a time. The call is free, it runs 30 minutes, and there is nothing to prepare.

Tell us where your team is losing time. You don't need to know what the solution looks like, and you don't need to have decided on anything.

The first thing anyone pays for is Discovery, and it comes after the call, not before it.

Every message is read by the person who would do the work, and you get a reply within two business days either way.

Or email hello@anum.com.au directly.
We don't sell your details or pass them to anyone for marketing. No CRMs. No mailing lists.
Anum Pty Ltd collects what you enter here to reply to your enquiry and for nothing else. It is held by our website host in the United States and reaches us in a Google mailbox. The providers, the retention period and how to get a copy or have it deleted are in our privacy policy.
Most engagements start with Discovery, priced in writing before it starts, not a proposal.

Message received

Thanks. It has landed and it will be read properly. You will hear back within two business days.

Where the call leads

If you go ahead, Discovery is the first paid step. This is what it hands you.

  • A written opportunity map: every workflow watched beside the people doing it, then ranked by value and effort.
  • A recommendation on what to do first.
  • The baseline: the two or three numbers the Rollout would have to move, measured before anything changes.
  • A do-not-automate list, and why each item is on it.
  • The governance gaps: what an AI use and data handling policy for your business has to cover.
  • A readback session, where we walk you through what we found before any of it is written up.
  • A training session with your team, built from what we found: getting real work out of the AI you are already paying for, and what can and cannot safely go near client and staff information.
  • A Rollout proposal, scoped and priced, in writing.

More questions? The ones that come up on the call are answered here.