Welcome back to Neural Notes, a weekly column where I look at how AI is influencing Australia. In this edition: why small businesses should actually be using less AI. Related Article Block Placeholder Article ID: 335397 AI usage surging in Australia’s online gambling sector David Adams Everywhere small business owners look right now, they’re being told that if they’re not all-in on AI, they’re already behind. Vendors, consultants, and LinkedIn ‘thought leaders’ insist that more tools, more agents and more automation are the only way to stay competitive. But for a lot of small teams, the reality isn’t that simple. Or accurate. Just over one-third of SMEs in Australia are now using AI, according to the government’s own figures. In regional areas, that drops to closer to 29%. But unlike the big end of town, they don’t tend to be building sleek AI-powered workflows. They’re juggling a chaotic pile of logins, browser tabs and half-adopted copilots. Rather than a strategic AI roadmap, it’s more akin to an AI patchwork quilt. Smarter business news. Straight to your inbox. For startup founders, small businesses and leaders. Build sharper instincts and better strategy by learning from Australia’s smartest business minds. Sign up for free. * indicates required Email Address * By continuing, you agree to our Terms & Conditions and Privacy Policy. What if instead of listening to the particular brand of AI bros that began peddling their “GEO” courses after pivoting from crypto, we tried something else instead? Less AI. On purpose. Approaching AI from a minimalist standpoint doesn’t mean rejecting automation. Instead, it’s a deliberate decision to keep your stack small, the use cases narrow, and the guardrails airtight.  The hidden cost of AI tool sprawl for small businesses If we’re being honest, for a small company, there is only so much complexity and stitched-together “solutions” that can be absorbed before the benefits begin to evaporate. One person signs up for a generative writing tool because social posts are a pain. Another experiments with a design app. Someone else starts using the assistant in the CRM. A vendor bundles an AI agent for support tickets. Meanwhile, the office suite quietly switches on its own sidebar assistant. And all of these decisions make sense in theory, particularly for a small operation low on time. Efficiency is king.  At first, the monthly fees feel small, and every tool is sold as a productivity gain. But if you look at the picture as a whole, what you see is tool sprawl: overlapping products doing similar things, with no one owning the system end-to-end.  Related Article Block Placeholder Article ID: 333633 Neural Notes: Lee Hickin on why Australia’s AI edge isn’t what you think Tegan Jones Then there are the hours spent setting each tool up, the time staff spend learning another interface. Some people half-use the tool and half fall back to old workflows.  What you can end up with in one small team is three different AI tools that may be used to write the same client content — none integrated, each producing slightly different outputs, and all requiring manual checks. Multiply all of that across every new product, and the returns start to look less compelling. Where’s the productivity gain in that? Every extra tool means another onboarding, another login, another place client data might end up, another set of quirks everyone has to learn by trial and error. No one is paid to manage this properly, so it gets handled ad hoc, if at all. AI minimalism starts by facing this reality. Instead of asking what else you could automate, what about taking a look at which two or three tools are actually worth the cognitive load.  Why AI adoption risks are rising for SMEs And then there is the question of risk. Most small businesses using AI do not have a written policy or a clear set of rules about what staff can paste into tools, or when a human needs to sign off on AI-generated work. This isn’t criticism. Small businesses are busy and that’s a big task.  Related Article Block Placeholder Article ID: 332999 Neural Notes: Anthropic data shows how AI is actually showing up at work Tegan Jones The governance conversation has moved faster than most founders have time to follow. At the same time, the marketing around AI tends to gloss over the basics: data handling, privacy, copyright and the very real possibility that these systems will simply make things up. The result is a messy status quo. Employees are encouraged to experiment, often in consumer-grade tools. Customer details end up in prompts. Internal documents get pasted into chatbots for quick summaries. AI-generated content goes straight to websites and social feeds, complete with hallucinated facts or unattributed borrowings. Not to mention those who have plugged their entire work and personal lives into the likes of OpenClaw. A security nightmare waiting to happen. For most small businesses, I’d argue the real risk is not that AI doesn’t work, but that it works just well enough to be trusted and go unchecked.  Individually, each of these risks might look manageable. Together, they create exposure that many small businesses just aren’t equipped to handle, whether it is a client dispute, a compliance issue, or reputational damage. It is not realistic to expect a 12-person company to build a full-scale governance framework. What is realistic is to do something more modest. Keep the number of tools low, understand how each one handles data, and set clear rules that people actually follow. This is not an argument against AI, or against the platforms building it. In many cases, the most effective approach for small businesses is to rely on a small number of trusted systems or a single vendor, where integrated features are often the most practical place to start. Teams only have so much bandwidth. So if you focus on a few high-value use cases, such as the work that is repetitive, time-consuming, or genuinely “good enough”, you are more likely to see real gains. In practice, this requires restraint. Audit the AI tools your team already uses. Set a cap on your stack and treat that as a constraint. Favour AI that sits inside platforms you already rely on. Write a one-page policy that makes it clear what is and is not acceptable. Choose a small number of measurable outcomes and be willing to walk away if they do not stack up. None of this will get you on a conference stage talking about your cutting-edge AI strategy. And thank god, because we really don’t need any more of those charlatans. Stay in the know Never miss a story: sign up to SmartCompany’s free daily newsletter and find our best stories on LinkedIn.