AIat workStart an enquiryEnquire

Learn to use AI on the work you're actually behind on

Practical AI training for Australian practices, agencies, clinics, firms, providers and trade businesses. Built around the admin and operational work that fills your week, not a generic prompt list.

We teach Claude and ChatGPT side by side, so you learn a skill that survives the next tool. We don't sell either one, and we'll tell you when the free version is enough.

Two tools, taught side by side

Most courses teach one tool and quietly become a sales channel for it. We teach Claude and ChatGPT together, because the useful skill is knowing which one to reach for and what each does with your information.

Where your data goes differs between vendors and between tiers. That difference is the first thing a practice principal needs answered, so it's the first thing we cover — not a disclaimer at the end.

Claude

Long documents, sustained drafting in your own voice, and thinking through something complicated without losing the thread.

ChatGPT

Fast turnaround at volume, voice notes into finished drafts, and the quick jobs you do twenty times a day.

And when to use neither

Some work doesn't belong in a general-purpose AI tool at all. Knowing where that line sits is part of the training, not a footnote to it.

Four people are in the room

Every team we walk into contains someone quietly worried about their job, someone who has never opened one of these tools and won't say so, someone teaching themselves in a free account, and someone who has to decide for all three. Training that answers only the software question loses three of the four before lunch.

Four people are in the roomA two-by-two figure showing the four attitudes present in every team: someone worried about their job, someone who has never opened an AI tool, someone already teaching themselves, and the person who decides for all three.Who is actually in the roomFour people are in the roomTraining that answers only the software question loses three of them before lunch.01Worried about their jobHas read the headlines. Will be polite in thesession and change nothing after it.→ Answer the replacement question first, out loud.02Never opened oneWon't say so in front of the team. Needs amental model before a prompt.→ One job, start to finish, on their own laptop.03Already teaching themselvesUsing a free account, unasked. Fastest in theroom and the biggest exposure.→ Close the three gaps self-teaching leaves.04Has to decide for all threeOwns the P&L and carries the risk. Wants therule, the cost and the boundary.→ A written policy, then training that defends it.ai-at-work.au — the position behind all four: this is about the team you already have getting more done.
Each of the four needs a different thing from the first hour, and the session is sequenced around that.

The position behind all four, stated once: this is about the team you have getting more done. Why that matters to the session.

How it works

Three steps, and the first one is free and takes about four minutes.

Answer nine questions

How the business runs, what's eating the week, what's already happening unofficially, and whether there's a written rule. No account, no scoring theatre.

Get a plan, and an honest read

Modules ordered for your work, plus where you're exposed right now. Sometimes the answer is that training isn't the first thing you need.

Decide whether to bring us in

A half-day with the team, a workflow built with you, or a policy and audit first. Nothing about the plan requires you to.

Why the industry page matters more than the course

The tasks people need AI training for are roughly the same everywhere: drafting, summarising long documents, structuring messy information, client communication, reusable setups, and knowing when not to use it at all. What differs is the vocabulary, the deadline that hurts, and the line you cannot cross without a problem.

So there is one core skill set, written once, taught properly — and a thin application layer per industry that makes it land. A practice owner should read three lines of their industry page and recognise their own Tuesday. If they don't, that page isn't ready and we haven't published it.

That also means we say no to some work. Anything where a wrong output is a regulated harm rather than an inconvenience — clinical decisions, financial advice, tax positions, safety sign-off — stays with the qualified person, and the training says so out loud.

What a session actually looks like

Not a lecture with a slide deck. About two thirds of the time is your team doing their own work while someone who knows the tools watches over their shoulder and corrects the habits.

Half a day, as it actually runs

  1. 0:00Where information goesWhich tool, which account, what never goes in, and what stays with a qualified person — using your documents as the examples.
  2. 0:40One real job, start to finishA job from your business, taken from blank to something you would send.
  3. 1:45Everyone does it themselvesOwn hands, own laptop. Where the habits get corrected.
  4. 3:00Setups you keepTwo or three saved setups on your most common documents. This is the part that's still there in a month.
A two-hour session covers the first two blocks and the last one. The hands-on middle is what a half day buys.

What's different three weeks later

The test isn't the feedback form at the end of the day. It's one question later: which job is now being done differently, and by whom? One named job, done differently by more than one person, is a session that worked.

The jobHow it runs nowHow it runs after
The standard explanatory letterRewritten from scratch, or copied from an old one and half-edited.A saved setup produces your version from three or four facts, in your voice.
A long document you need the gist ofRead in full by whoever has time, usually the most expensive person available.Answered against your own checklist, with the source open beside it.
The monthly reportAn evening, often a weekend, done by the owner.A structured draft in an hour. The judgement stays yours.

Tell us which of these is your version and we'll tell you which engagement fits.

Start an enquiry
What is different three weeks laterThree named jobs, showing how each runs today and how the same job runs after training, with the judgement staying with the person.The only test that countsWhat’s different three weeks laterNot the feedback form. One question: which job is now being done differently, and by whom?The jobHow it runs nowHow it runs afterThe standard explanatoryletterRewritten from scratch, or copied froman old one and half-edited.A saved setup produces your versionfrom three or four facts, in yourvoice.A long document you needthe gist ofRead in full by whoever has time —usually the most expensive personavailable.Answered against your own checklist,with the source open beside it.The monthly reportAn evening, often a weekend, done bythe owner.A structured draft in an hour. Thejudgement stays yours.One named job, done differently by more than one person, is a session that worked.
One named job, done differently by more than one person, is a session that worked.

The boundary, drawn

Two columns, and the right-hand one doesn't move with the tool, the tier or the vendor. Tenders, policies, induction material, standard explanatory letters and reports from de-identified inputs sit on the left. Clinical judgement, financial advice, tax positions and safety sign-off sit on the right, permanently.

We also teach both Claude and ChatGPT with no reseller agreement, partner status or referral fee in either direction, and we'll tell you when a free tier is enough. The reasoning is on why we teach two tools.

The boundary, drawnTwo columns. On the left, work that can safely be drafted with an AI tool. On the right, work that stays with a qualified person regardless of tool, tier or vendor.The line you cannot crossThe boundary, drawnThe right-hand column does not move with the tool, the tier or the vendor.Draft it with AIYou still check it. You always check it.Stays with a qualified personPermanently. Not a tier or a setting away.Tenders and submissionsPolicies and proceduresInduction and training materialStandard explanatory lettersReports from de-identified inputsFirst drafts of anything routineClinical judgementFinancial adviceTax positionsSafety sign-offLegal advice to a clientAnything a regulator would call your opinionA wrong output on the left costs you an edit. On the right it is a regulated harm.
A wrong output on the left costs you an edit. On the right it is a regulated harm.

Questions we get before the first call

Who is this for?

The person who owns the P&L and carries the risk — an owner, principal, practice or office manager, or director. Staff are in the room; the decision isn't theirs to make.

Do we need to know anything already?

No. The normal room is half cautious, half already using it unofficially, and the sequence is built for exactly that mix.

What if our work is regulated?

Then the written rule comes before any training, and that's a separate piece of work. See policy and audit.

Is there anything to read before enquiring?

Plenty, and none of it is gated. The guides and the blog carry the reasoning we'd give you on a call.

Get a training plan built for your business. Nine questions about how your business actually runs. You get a plan ordered for your work, a reading on where you're exposed, and one clear place to start. No account needed. Build your plan.

Start an enquiry

Tell us what's eating the week

One form, one reply from a person. We'll tell you which engagement fits, roughly what it costs, and if we're the wrong fit we'll say so.

Start an enquiry
Start an enquiry