AI August 2026

I run a small business, so I have no interest in adding to the AI hype. What follows is not a prediction or a pitch. It is what a year of actually using these tools, on my own business, has taught me, with apologies to the old Western: the good, the bad, and the bot.

The good

The honest starting point is that AI got me unstuck. I am the kind of person who can start ten things and finish three, so momentum, not motivation, was always my problem. AI removed the blank page. "Draft this." "Give me a first version of that." Work that used to stall for days was suddenly moving, and I was editing and adding my own judgement instead of staring at nothing.

The bigger shift was mental. I stopped treating AI as a helper inside Word or Excel, like the old Clippy, and started treating it as the thing that drives those tools for me. "I'm in Excel, help me" became "I want to build this, what is the best way." Once that clicked, the results surprised me. Over a weekend I built a live performance dashboard that pulls our previously separate systems, our helpdesk, CRM, quoting and device management, into one read-only view. We rebuilt this entire website, which we had started and abandoned two or three times over the years. We added tools I would never have attempted alone: a three-minute security health check, a cost estimator, a hardware replacement calculator.

I am not a developer. What AI gave me was not skills I lack, it was the ability to finish the things I could always picture but never build. Used well, it removes the bottleneck between knowing what you want and actually having it.

The bad

Now the honest part. The biggest risk with AI is not that it fails, it is that it succeeds too easily. It will cheerfully tell you "yes, I can do that," and before long you feel like Superman, which is exactly when you get into trouble. Two things bite people.

First, it agrees with you far too much. Ask a leading question and it will happily confirm whatever you assumed, so you have to push back and treat its confidence as a starting point, not an answer. Second, it does not know what good looks like, and it assumes you do. When I use AI to draft an incident response plan, thirty years in IT means I can glance at it and say "that is wrong, change this." Someone who has never handled a real incident cannot. We already see clients open a support ticket and paste in the fix ChatGPT gave them, and often there is something useful in there, but knowing which part to trust is the entire job. It is also not free once you rely on it, so plan for real usage limits and monthly cost rather than being surprised by them.

The bot

Which brings me to the bot itself. The AI that built that dashboard, this website and those calculators is genuinely strange to work with, in ways my old mental models did not prepare me for. It went and found its own research studies to justify a calculator I was building, some of which I had to rein in. It rebuilt our website not with a modern content system but as clean, plain HTML, effectively winding web development back twenty-five years, and somehow that was the right call. It built tools I had only half-imagined. At one point it calmly announced it was updating all seventy-seven pages of our site in one go, and I genuinely paused and thought, is that too much? And it does what you ask and, mostly, does not argue.

Here is the honest reveal. The site you are reading, and the tools on it, were built by the bot with me sitting beside it. That is not a gimmick or a shortcut, it is the whole point. A capable AI does not turn me into a web developer. It lets someone who knows exactly what they want, and what good looks like, finally get it built. The bot does the building. I still have to know what to build, and when to say stop.

What it actually means for your business

Here is where I land. AI is a remarkable accelerator. It removes the grunt work, gets you unstuck, and lets people who know what they want build things they never could before. What it does not do is replace judgement. The skill that matters now is not prompting, it is knowing where to stop: what to let it touch, what to keep read-only, what to check before you trust it, and what data should never go near it in the first place.

Treat it as an accelerator with guard-rails, not an autopilot. That is how we use it ourselves, and how we help our clients adopt it through AgileAI: deliberately and safely, in a way that fits their business rather than the other way around. The technology is genuinely remarkable. The judgement about where to use it, and where to stop, is still a human job. That is the part we are here for.

If you want a straight, hype-free conversation about where AI could genuinely help your business, and where it should not, that is one we are always happy to have.

Frequently asked questions

Will AI replace developers, or IT professionals?

Not in our experience. AI does not give you skills you do not have, it lets people who already know what they want, and what good looks like, finally build it. You still need the judgement to know what to build, what to trust in the output, and when to stop. The tool does the building, the human does the deciding.

What is the biggest risk of using AI in a small business?

The biggest risk is not that AI fails, it is that it succeeds too easily and encourages you to do things you should not. It will happily agree with you and confidently do things it cannot verify. The safeguard is judgement: start read-only, give a tool access to one specific job rather than the keys to everything, check the output before you trust it, and keep sensitive data away from it entirely.

Is it safe to use AI on business data?

It can be, with the right care. The critical point is protecting not just your own data but your customers' data, your financial information, and any information you hold on behalf of other people. We work deliberately: read-only first, scoped access to specific tasks, private previews before anything goes live, and never pushing data somewhere we are not certain is secure just because a tool said it could.

Did Agile IT really build its website with AI?

Yes. This website, along with a performance dashboard, a three-minute security health check and several calculators, was built with an AI assistant working alongside our Managing Director. It was written as clean HTML with no content management system, and it is a working example of using AI with a clear idea of the result you want and firm guard-rails around what it can touch.

Where should a small business start with AI?

Start small and reviewable. Pick one contained task where you already know what a good result looks like, keep a person checking every output, and expand only once you trust it. Treat AI as an accelerator with guard-rails, not an autopilot. If you want a hype-free conversation about where it could genuinely help, that is what an AI Discovery session is for.