AI Fast TrackerShould You Build Your Own Tool, or Buy One? An Australian Small Business Guide (2026)
There is a moment most small businesses hit. The off-the-shelf software does nine of the ten things you need, the tenth one is the one that actually costs you time, and the quote to have something custom built comes back at four figures. In 2026 there is a third option that did not really exist before, and it is worth understanding before you spend anything.
What actually changed
Until recently, building a small internal tool meant either learning to code or paying someone who already had. Both are expensive in the currency you have least of.
What changed is that you can now describe what you want in plain English and get a working application back, running on the internet, that you can hand a link to. Two things made that possible at once: models good enough to write and fix real code, and hosting platforms that handle everything a working application needs without you touching a server.
Replit is one of those platforms and probably the most direct example. You describe the thing you want, it writes it, runs it, and gives you a URL. Claude and similar models do the writing. You stay in plain English the whole way.
What this is genuinely good at
Be specific about the shape of problem it solves, because the honest answer is narrower than the marketing suggests.
- The internal tool nobody sells. A calculator that uses your actual pricing rules. A form that turns a job enquiry into the exact quote format you already use. A dashboard that shows the four numbers you care about. These are too specific for anyone to build as a product, which is why you have never found one.
- The thing you have been doing in a spreadsheet for years that half your team gets wrong because the spreadsheet has no rules in it.
- A prototype to think with. Sometimes you do not know what you want until you use a wrong version of it. Building a rough one in an afternoon is cheaper than specifying the right one in a meeting.
When you should not do this, and this is the important part
Most guides stop at the first section. These are the situations where building your own is the wrong call, and recognising yourself in one of them will save you real money.
- Someone already sells it, well, for under a hundred dollars a month. Invoicing, accounting, scheduling, CRM. These are solved. Building your own version of Xero is a hobby, not a business decision.
- It has to hold customer data and you have not thought about privacy. If a tool stores your customers' personal information, you have obligations under the Privacy Act, and building it yourself does not reduce them. Note that since 1 July 2026, businesses providing services designated under the AML/CTF Act, which includes real estate agents, accountants, lawyers and conveyancers, must comply with the Privacy Act for that information regardless of turnover. Check your position at oaic.gov.au.
- It has to take payments. Use a payment provider's own checkout. Do not build this.
- Nobody in the business will maintain it. A tool with no owner becomes a liability the first time it breaks, and it will break the week you are busiest.
- You want to sell it. A working prototype and a product you can support are very different things, and the distance between them is where most of the work lives.
The cost, honestly
This is where people get caught, so read it carefully.
Replit's pricing is usage based. You are not buying a flat monthly seat; you are paying for the work the machine does on your behalf, and a session where you go back and forth many times costs more than one where you get it right early. Several independent write-ups have criticised exactly this, because the bill is hard to predict before you start.
What to do about it. Set a spend limit in the platform before your first real session, not after. Treat your first build as a fixed experiment with a number attached that you have decided in advance you are willing to lose. If you would not spend that on a tradesman's call-out fee to look at the same problem, do not spend it here either.
There is a free tier, which is enough to find out whether the idea works at all. Confirm the current plans and limits at replit.com before you commit, because pricing on these platforms changes often and anything written here has a shelf life.
How a first build actually goes
Not a tutorial, because the interface will have changed by the time you read this. The shape of it, which does not change:
- Describe the job, not the software. "A page where I type a job type, square metres and travel distance, and it gives me a quote using my rates" works far better than a list of technical requirements.
- Give it your real rules. Your actual pricing, your actual minimum call-out, the discount you actually give repeat customers. Generic input produces a generic tool you will not use.
- Use it before you improve it. Run three real jobs through the rough version. You will discover the thing you forgot, and you will discover it in five minutes rather than after a week of polishing.
- Stop earlier than feels right. The tool needs to be better than the current manual process. It does not need to be good software.
The honest summary
Building your own small tool went from a project to an afternoon, and that is a real change worth knowing about. It is not the answer to most problems. It is a very good answer to the narrow, specific, nobody-sells-this problem that has been quietly costing you an hour a week for years.
If you have one of those in mind right now, that is the one to try. If you had to go looking for a reason to use it, that is your answer.