What AI Can Actually Do for Your Business (And What It Can't)

I need to be honest with you about something. There’s a lot of nonsense being said about AI right now.

On one side, you’ve got people telling you AI will replace every employee and run your entire business by Tuesday. On the other, people insisting it’s all smoke and mirrors and nothing works. Both are wrong, and both are costing business owners real money — either by chasing magic solutions that don’t exist or by ignoring genuine advantages that do.

So here’s my honest take, from someone who uses AI every single day to build real things for real businesses.

What AI Is Genuinely Great At

Understanding legacy systems. This is AI’s superpower that nobody talks about. Hand it a codebase that’s been patched and modified for a decade, and it will map every connection, every dependency, every edge case. Work that used to take a senior developer two weeks of careful archaeology happens in hours.

Consistent quality. AI doesn’t have bad days. It doesn’t get tired on a Friday afternoon and miss a security header. It doesn’t forget to add input validation because it was rushing to meet a deadline. Every output follows best practices because it literally cannot forget them.

Speed of execution. Once the strategy is set and the direction is clear, AI can build at a pace that fundamentally changes project economics. I’m delivering in days what used to take weeks. That’s not an exaggeration — it’s measurable.

Breadth of knowledge. No single developer knows every framework, every language, every security standard. AI does. It can work across technologies without the learning curve that used to eat up project time and budget.

Handling repetitive precision work. Setting up proper SEO tags across fifty pages. Ensuring consistent security headers. Writing tests for edge cases. This is work that’s important but mind-numbing for humans. AI handles it without the quality drop-off that comes with boredom.

Where AI Falls Short

And here’s where I lose people who are trying to sell you an AI fairy tale.

AI doesn’t know your business. It can build anything you describe, but it can’t tell you what to build. It doesn’t know that your customers in Geelong respond differently to messaging than your customers in Melbourne. It doesn’t know that your biggest competitor just launched a feature you need to respond to. It doesn’t know your revenue mix or your growth plans.

AI doesn’t replace strategic thinking. Should you rebuild your website or fix the existing one? Should you invest in a customer portal or a better booking system? Which features will actually drive revenue? These are human decisions that require understanding context AI simply doesn’t have.

AI can be confidently wrong. This is the one that catches people. AI will give you an answer with complete confidence, and sometimes that answer is incorrect. Without someone who knows what right looks like, you won’t catch it. And wrong code deployed to production is worse than no code at all.

AI doesn’t handle truly novel problems well. If nobody has ever solved a problem similar to yours, AI doesn’t have patterns to draw from. It’s exceptional at combining existing solutions in new ways, but genuinely unprecedented challenges still need human creativity.

AI doesn’t manage relationships. It won’t call your hosting provider when something goes wrong at midnight. It won’t negotiate with your payment processor. It won’t sit in a meeting and read the room to understand what your stakeholders actually want versus what they’re saying.

The Combination That Actually Works

Here’s what I’ve found after two decades of building things and the last stretch of using AI intensively: the magic isn’t AI alone. It’s AI capability combined with human judgment.

I use AI to handle the heavy lifting — the code generation, the security implementation, the testing, the repetitive precision work. That frees up my time to focus on what AI can’t do: understanding your specific business situation, making strategic decisions about what to build, catching the moments where AI’s suggestion looks right but isn’t.

The result is better than either could produce alone. Faster and more consistent than a human working solo. More strategically sound than AI working without experienced oversight.

What This Means Practically

If someone’s selling you “AI-powered” anything, here are the questions to ask:

Who’s making the strategic decisions? If the answer is “the AI,” walk away. AI should be executing strategy, not setting it.

What happens when something goes wrong? AI-generated code needs human oversight. A developer with real experience needs to review what’s being built, catch problems before they reach your customers, and make judgment calls AI can’t make.

Is the timeline realistic? AI is fast, but “we’ll rebuild your entire business in a weekend” is still fantasy. Days instead of months is real. Hours instead of weeks is real. Instant transformation is not.

What experience backs up the AI? AI is a tool. A powerful one, but still a tool. It’s only as good as the experience guiding it. Twenty years of building software means I’ve seen what works, what fails, and why — and AI amplifies that experience rather than replacing it.

The Bottom Line

AI genuinely changed the economics of software development for small business. Projects that were out of reach are now affordable. Timelines that were impractical are now realistic. Quality that required expensive teams is now achievable at a fraction of the cost.

But it’s not magic. It’s a powerful tool wielded by experienced hands.

If AI can do something for your business, I’ll tell you exactly what and how. And if it can’t? I’ll tell you that too. Straight talk is more valuable than a sales pitch, and it’s the only way I know how to work.