Skip to content

The Experiment

The experiment, day 0: an AI is building this website

We are testing whether AI can build, launch, rank and maintain a real commercial website end-to-end. This is the day-0 baseline, the rules, and the public targets we will be judged against.

A lab notebook open beside an ascending line chart, progress being recorded by hand

Today is day 0. If you’re reading this, an AI wrote it, built the page it sits on, and published it — supervised by me, but built by the machine. This post is the baseline we’ll be measured against.

The thesis we’re testing

AI can build, launch, rank, and maintain a commercial website end-to-end — fast enough and well enough that an “AI-managed growth engine” becomes a product you can actually sell.

This site, ai4smb.com.au, is demo unit #1. It’s not a mock-up. It’s a real business — an AI automation agency for Australian small business — with real offers and a real flagship product (an AI receptionist). The experiment and the business are the same thing on purpose. If the AI can rank this, in a competitive category, from nothing, that’s the proof.

The baseline (be honest about the starting line)

As of today:

  • Not indexed for our own brand term. Google doesn’t know we exist.
  • Zero authority. No backlinks, no history, no trust signals.
  • Zero content until this launch set went live.

A genuine clean slate. No head start to hide behind, which is the point — a clean slate is the only fair test.

The rules

  1. The AI does the work. Building, copywriting, SEO structure, the content engine going forward — all AI-driven. A human (me, 30 years in IT) supervises and approves, especially early. Autonomy gets dialled up as the thing earns trust.
  2. We publish results either way. Good numbers or embarrassing ones, the day-30, day-60 and day-90 reports go up. An experiment you only report when it works isn’t an experiment; it’s an ad.
  3. No fake authority. No bought links, no PBNs, no tricks. If it ranks, it ranks on content and legitimate signals.
  4. Build time counts. How fast AI can go from empty repo to launched site is itself a metric. The clock started today.

The public scorecard

The targets we’ve set ourselves, tracked live on the experiment page:

  • Week 1 — live, indexed, Search Console verified, rank tracking active.
  • Day 30 — top 20 for our low-difficulty targets; top 3 in Australia for branded “ai4smb”.
  • Day 60 — top 10 for “ai receptionist australia”; first organic enquiry.
  • Day 90 — top 5 for “ai receptionist australia”; top 10 for five or more money terms; cited in at least one AI engine answer.

Why do this in public

Because it’s also the product. The top tier of our managed services is an AI-managed content and SEO engine — the exact capability building and maintaining this site. The most honest sales demo I can think of is to point it at my own business, set public targets, and show you the real numbers as they land.

So: bookmark the experiment page. Day 30 is where it gets interesting. Win or lose, you’ll see the data.

Not sure where AI actually pays back?

Start with an AI Opportunity Audit. You get a prioritised map of where automation earns its keep in your business — and a roadmap you own, whether you build it with us or not.