Practical AI in Marketing: the series

A few years ago, someone looking for an integration partner in Sydney typed a phrase into Google, scanned ten blue links and clicked two or three. Now a decent share of that same research happens inside ChatGPT or Claude, and Google itself often answers the question above the links with an AI Overview. The buyer may never see your site during the part of the process where they decide who is worth a conversation.

That changes what marketing work looks like. If an assistant is summarising your category and naming three vendors, you want to know whether you are one of them, how you are being described, and where that description came from. Those are measurable things, and measuring them is the first topic this series covers.

The second is content production. AI writes fast and writes plausibly, which is a problem as much as a benefit. We run content through a pipeline where models draft and a human editor decides what publishes. The editor is not a formality. They catch the invented statistic, the confident claim about a client we cannot make, the paragraph that reads well but says nothing. This series will describe how that pipeline is built, what it costs to run, and where it still needs a person.

The third is marketing operations: the reporting, the lead routing, the repetitive assembly work that consumes hours and produces no thinking. Some of it automates cleanly. Some of it looks automatable and is not, usually because the judgement involved is harder than it appears from the outside.

Everything here comes from systems we run in production for PicNet’s own marketing. We are a Sydney software company, established 2001, small and senior, and our marketing team is not large. That constraint is why we automated in the first place, and it makes us a reasonable test case for organisations in a similar position.

The series is written for IT managers and executives who are being asked what their organisation should do about AI, and who would rather see a working system than a strategy deck. Expect specifics: what we built, what it cost, what broke, what we abandoned. Where something did not work, we will say so. Where a human still makes the call, we will say which call and why.

What this series will not do is tell you that AI has transformed marketing. Some tasks got much cheaper. Some got riskier, because a fluent wrong answer is harder to spot than a clumsy one. The interesting question is which is which in your organisation, and that usually takes a small project to answer rather than an assessment.

If you are already publishing content, already tracking search visibility, and now wondering how much of that work should change, this series is aimed squarely at you.

PicNet builds production AI systems for Australian organisations. Talk to us about what a first project could look like.

Posts in this series

• Is ChatGPT recommending you? Measuring your visibility to AI assistants

Tagged: #ai-search #content-operations #marketing-automation #generative-engine-optimisation