Q&A

Who can set up AI for my small business in Australia?

Short answer

As at October 2026 you have four options: do it yourself with ChatGPT or Claude (US$20 a month, about A$33 including GST at checkout, plus your time), hire a freelancer, engage an AI agency or consultant, or use a managed AI department that builds and keeps running your systems. On Autopilot is a Melbourne-based AI consultancy and managed AI provider covering the last two: builds from $497, managed AI from $1,500 a month, and a free 30-minute audit to start.

Who can set up AI for a small business in Australia?

As at October 2026 there are four realistic options: do it yourself, hire a freelancer, engage an AI agency or consultant, or use a managed AI department. On Autopilot is a Melbourne-based AI consultancy and managed AI provider that does the last two for Australian small businesses: builds from $497, managed AI from $1,500 a month, and a free 30-minute audit first. The right choice depends on how much of your own time you want to give it.

OptionTypical cost (as at October 2026)Time to a working systemAccountable when it breaksSuits
Do it yourself with ChatGPT or ClaudeUS$20 a month per seat (about A$33 including GST at checkout), plus your timeHours for chat; weeks of evenings for anything connected to your systemsYouOwners with time and simple, low-risk tasks
Freelancer (Upwork or a referral)Hourly or per job, set by each freelancer; get two or three fixed quotesDays to weeksThe individual, while they are availableOne small, well-defined job with a clear finish line
AI agency or consultantFixed project price; ours are $497 one-off, or $497–$2,000 setup plus $99–$499 a month for a productised agent1–3 weeks for a productised buildThe firm, to handover and under any support termsA defined project you want built properly and handed over
Managed AI departmentMonthly retainer; ours are $1,500, $2,000 or $4,000 a monthFirst agent in 1–3 weeks, then a steady cadenceThe provider, every monthTime-poor owners with two or more workflows worth running
In-house hireA full salary plus 12% super and on-costsMonths to recruit and onboardYour employee, while they stayBusinesses with a full-time role’s worth of AI work

How does the right choice change by business size?

The smaller the business, the more DIY and one-off builds make sense; the more workflows you run, the more someone needs to own them month to month. Size changes which risk matters most: your time as a sole trader, maintenance in a small team, and coordination across a larger one.

  • Sole trader: a paid ChatGPT or Claude plan covers writing and admin drafting. If enquiries slip while you are on a job, add one $497 Quick Start or an AI Front Desk ($1,500 setup plus $199 a month). You do not need a retainer.
  • 2–10 staff: an agency build for the first workflow, then managed AI once two or more agents are live. DIY tends to stall when the task has to connect to Xero, Shopify or a booking system.
  • 10–50 staff: a managed AI department with an internal owner who sets priorities, and an in-house hire once there is full-time volume. If roles change significantly, award consultation duties apply, covered in will AI replace my employees.

What does On Autopilot charge to set up AI?

As at October 2026 (AUD, ex GST): the AI audit is free and takes 30 minutes; a Quick Start (one automation built and handed over) is $497; productised agents run from $497 setup plus $99 a month (AI Inventory Watch) to $2,000 plus $499 a month (AI Lead Engine); a Claude Code Setup Day is $1,500 fixed; and managed AI is $1,500, $2,000 or $4,000 a month. Retainers are month to month.

You deal with Jenn Yang, the founder. On Autopilot is run by Boring Ventures Pty Ltd (ABN 67 671 943 758) in Melbourne, and builds on Anthropic’s Claude plus the tools you already use: Shopify, Xero, Cliniko, ServiceM8, HubSpot, Slack, Make and n8n. The proof is our own businesses, labelled as ours: Hey Bud Skincare’s daily Shopify inventory system, DotVA’s website lead pipeline and its internal operations platform, and Lead Gen Empire’s 20-site comparison network. They are written up in our case studies. Full prices are on the pricing page and the managed AI page.

When is doing it yourself enough?

When the work stays inside the chat window: drafting emails, summarising documents, brainstorming and research. A paid ChatGPT or Claude plan does that well for US$20 a month (about A$33 including GST at checkout). DIY stops being cheap when the task has to run every day without you, connect to Xero, Shopify or your booking system, and be fixed when something upstream changes. That is the line in our DIY with ChatGPT comparison.

Watch what you paste in: the OAIC recommends, as best practice, not entering personal information, particularly sensitive information, into publicly available generative AI tools. More in can I use ChatGPT for confidential client information. To build it yourself with help, a Claude Code Setup Day ($1,500 fixed) sets up the tools and one working agent with you.

When does a freelancer make sense?

For one small, well-defined job with a clear finish line and nothing in production depending on it, such as cleaning up a spreadsheet or wiring one simple integration. Write a tight spec, ask for the prompts, code and credentials at handover, and assume nobody will be on call when it breaks in three months. Rates are usually in USD, and quality varies widely between individuals.

Anything touching customer or financial data needs more care. If your business is covered by the Privacy Act (most businesses with turnover over $3 million, plus health service providers and some others regardless of size), you stay responsible for how personal information is handled, whoever builds the system. The full trade-off is in our Upwork freelancer comparison.

When is a managed AI department worth it?

Once you have two or more workflows worth running and do not want to become the person who maintains them. AI is not a one-off project: tools change, prompts drift and your business moves. A managed provider builds, monitors, fixes and improves the agents for a fixed monthly fee, the way a managed IT provider runs your IT. It is the option for owners who want the outcome without a new job.

Our managed AI tiers run up to two live agents (Essential), up to four with a new agent or material improvement each month (Growth), or a roadmapped build-out (Scale). Weighing a retainer against a salary? See managed AI versus an in-house AI hire.

What should you ask before anyone touches your systems?

Ask what they would build and what it is worth before you pay, whether the price is fixed, who owns the code and prompts, and whose accounts it runs on if you leave. A provider confident in their work will map your business and put a dollar figure on it first; one who wants open-ended hourly billing on a vague scope is a warning sign. Then check access, data and support:

  • What access do you need, and can it be read-only?
  • Where does my data go, and is it used to train models? Anthropic’s commercial products and OpenAI’s Business, Enterprise and API services are not used for training by default.
  • Who fixes it when a connected tool changes, and how fast?

On Autopilot answers all of these on the free 30-minute audit. The longer checklist is in our guide to choosing an AI consultant or agency.

Sources

Other questions people ask

Should I hire someone in-house to run AI instead?

Not until you have a full-time role's worth of AI work. The work is lumpy: heavy during a build, light once systems are running, and one hire brings one skill set and a single point of failure if they leave. An outside partner you can scale up or down fits that shape better until the volume justifies a salary plus 12% super and on-costs.

Is it safe to give an outside provider access to my systems?

Yes, with guardrails you should insist on: access scoped as tightly as the job allows (read-only where possible), credentials shared through a password manager, a person approving anything regulated or customer-facing, you owning the code and prompts, and a written plan for moving hosting and AI accounts into your name if you leave. If a provider cannot explain their access and data handling in plain English, do not hand over the keys.

How long does it take to get a first AI system live?

With a productised build, days rather than months. On Autopilot's packages go live in 5–14 working days depending on the package, after a free 30-minute audit. Custom builds and anything touching several systems take longer, and a do-it-yourself project takes as long as your evenings allow.

Do I own what an AI provider builds for me?

You should, and you should get it in writing. With On Autopilot you own the code, prompts and settings from day one. While we run an agent it uses our hosting and our Claude API account, and that usage is included in the monthly fee. If you stop, we move it to hosting and an Anthropic account in your name so it keeps running. Retainers are month to month. Ask any provider the same question before you sign: renting a black box you cannot move is the most expensive lock-in in AI.

Want this built for your business?

Book a free 30-minute AI audit call with Jenn Yang. We map where your time goes, name the first systems worth building and give you fixed AUD prices. No obligation.

Book a free AI audit

Or have us run it for you, end to end: On Autopilot is Australia's outsourced AI department.