Why I built AI Workshift instead of becoming another AI consultant in Perth
AI Workshift builds virtual AI departments for small businesses in Perth, Western Australia. Each client gets named AI staff that run recurring work every week on infrastructure the client owns, sharing memory and handing jobs between each other. AI Workshift exists as an alternative to AI consulting, because advice on its own was never going to change a Tuesday morning.
I'm Bela. I run AI Workshift out of Perth. When I started this, the easy version of the job was sitting right there: print a card that says AI consultant, charge by the day, hand over a strategy document, move on to the next business.
I did not do that. This is why.
Why is Perth suddenly full of AI consultants?
Because the demand turned up before the people who can build anything did.
Every owner in this city has been told they are behind on AI. Some of that is noise. Some of it is fair. The ABS Business Characteristics Survey for 2024-25, published in June 2026, found 12% of Australian businesses reported using AI, up from 1% in 2021-22. Split it by size and it gets more honest. Around 11% of small and micro businesses with under 20 employees used AI, against roughly 35% of large businesses. The big end of town is three times further down the road.
So there is a real gap, owners can feel it, and a gap everyone can feel attracts people selling maps.
As of mid 2026 there are roughly 16 other AI providers operating in Perth. I have looked at nearly all of them. They sell tools, one off builds, or human consulting hours. That is not a scandal, and most of them are decent at what they do. It is just the easiest thing to sell. Advice has no infrastructure behind it, no support load, and no Monday morning where it has to actually work.
What do you actually get handed at the end of a consulting job?
A document. Usually a good one.
An audit of your systems. A list of opportunities ranked by effort and impact. A roadmap with phases. Maybe a workshop where the team gets genuinely excited for an afternoon.
Then the engagement ends. The document goes into Drive. Monday arrives and you are still writing quotes in the ute at 7pm, and the only person who knew how to build the thing is now at their next client.
I am not bagging out consultants. The advice is often correct. The model is what I have a problem with. Advice hands a plan to the one person who already has no time to execute it. Small business owners do not have a strategy shortage. They have a bodies shortage.
Deloitte Access Economics research commissioned by Amazon, published November 2025, makes the point better than I can, and it is vendor research so weigh it accordingly. Two thirds of Australian SMBs report using AI. Only 5% are what Deloitte calls fully enabled. Two thirds are having a fiddle. One in twenty has it running the work. Nothing in a slide deck closes that distance.
So what did I build instead?
Staff.
Not a chatbot. Not a dashboard. Named AI workers that do the recurring jobs of a small business every week, on a schedule, plugged into the systems the business already runs on. Email, calendar, CRM, Xero, the enquiry form on the website.
They share memory, so the one answering an enquiry knows what the one chasing the quote already said last Thursday. They hand work between each other, so an enquiry becomes a booking, becomes a job, becomes an invoice, becomes a polite chase if it goes unpaid. One chain, nobody drops the baton.
And I stay on to run them. That is the bit that makes it a department instead of a project. Businesses change. Prices change. You add a service, you drop a supplier, you start doing weekends. Somebody has to keep the staff current, and that somebody is me, here, not a ticket queue in another timezone.
Why a team rather than one clever bot is a longer argument, and I have written it out on why a team.
Who are Alex, Sarah, Marcus and Linda?
They are the four roles people ask for most, so they are where most builds start.
Alex runs front desk and booking. Answers enquiries, books jobs and appointments, sends confirmations and reminder SMS, and watches email and website enquiries around the clock.
Sarah is the account manager. She follows up every quote and every lead, keeps the CRM current, and flags the high value ones so you can make that call yourself.
Marcus does research and analysis. Weekly reports, KPIs, utilisation, where the bottleneck actually sat this month.
Linda handles books and compliance. Chases overdue invoices, reconciles in Xero, keeps the records audit ready.
Those four are a starting point, not a menu limit. I can build any agent for any role a business needs. A parts sourcing agent, a tender watcher, a roster planner, a warranty claims agent, something nobody has asked me for yet. The current line up is on the team page.
Isn't this just consulting with extra steps?
No, and the test takes one question. What is different about Tuesday morning?
After a consulting engagement, Tuesday morning is exactly the same as last Tuesday, plus a document you have not read since the handover meeting.
After a build like mine, Tuesday morning already happened while you were asleep. The enquiries that landed at 9pm are answered and booked. Last week's quotes got followed up. The 40 day invoice got a nudge. Monday's numbers are sitting in your inbox with the bottleneck circled.
There is thinking involved in what I do, obviously. You cannot build staff for a business you do not understand, so the first fortnight is a lot of questions about how you quote, who your good customers are, what you charge, what you keep forgetting to do. But the questions are the input. They are not the product. Nobody pays me for the questions.
And if you genuinely have spare capacity and no plan, hire a consultant. That is the right buy. Most owners I meet have the opposite problem.
What does it cost, and how fast does something go live?
Flat monthly tiers, published openly on the pricing page. No per seat games, no credits that run out halfway through a busy month.
"That sounds expensive" usually arrives before anyone has seen the number. Then people compare it to a part time admin hire and go quiet. Better comparison: what the current setup already costs you. Xero surveyed 500 Australian small business owners in early 2026 and found owners lose more than four hours of productive work a week to stress. Financial management alone eats around 21 hours a month. That is most of a working week gone, every month, to admin nobody enjoys doing.
First AI worker is typically live within two weeks. Not a pilot. Not a proof of concept. Working, on your systems, doing real jobs with real customers.
It all runs on infrastructure dedicated to your business, a private setup built for you rather than a shared product everyone else is also sitting inside. It is yours.
Built and supported in Perth, by me. If something goes wrong you get a person who knows your setup.
How do I know if I need staff instead of advice?
Three questions, and be honest with the answers.
Is there work you know should happen every week that reliably does not? Are you the bottleneck on things that need your time but not your brain? Would another set of hands change your month more than another customer would?
Two yeses and you have a capacity problem. A strategy document cannot fix a capacity problem. It can only describe one in more detail.
Have a look at the industries I work with and see if yours is there. If it is not, message me anyway. Most of the builds I am proudest of started with someone asking for a role that did not exist yet.
You do not need to be told what to do. You need the work to get done.
Top comments (0)