Last updated: 5 June 2026
I'd been putting off a website rebuild for over a year.
Every time I sent someone the link, I winced. The content was fine — that wasn't the problem. The problem was that I'd built myself a way of working with AI as a thinking partner, sitting next to me through every content and planning session. The more I leaned on it, the more it kept surfacing things about my own website that were off — a page taking a beat too long to load on mobile, a section that didn't quite line up, a footer link going to the wrong place, an analytics view I couldn't trust.
It came to a head after a particularly long planning session. The AI and I had spent the afternoon working through the next quarter's content — the angles, the calendar, the briefs — and at the end of it I sat back, looked at the site I was actually trying to drive traffic to, and realised it didn't deserve the work I was about to throw at it. The nagging "should I rebuild this?" question I'd been ignoring for a year became a real conversation with myself, that evening, with the same thinking partner I'd been using for the planning.
I ended up rebuilding the entire site over two days, with an AI as my thinking partner the whole way. Here's the honest version of how that went — and what I had to verify myself, because AI doesn't get to make the decision for you.
What "AI for Small Business" Actually Means — In One Paragraph
AI for small business isn't a software category you buy. It's a way of working — and one you have to be intentional about: using AI tools (the chat-style ones, mostly) as a thinking partner who can accelerate research, draft trade-offs, surface options you hadn't considered, and challenge assumptions — without ever being the one who decides. The decisions stay yours. The verification stays yours. AI is what lets a one-person business owner do the work of a small team's worth of thinking, but only if the owner stays in the loop and checks the work.
Why AI Changed How I Made the Decision
Before AI tools were good enough to be useful, "should I rebuild the site?" was the kind of question I'd put off for a year. I'd sit down to research it, read about hosting options, weigh the cost of rebuilding versus patching, scan forum threads — and I'd typically end the session with more questions than answers. Five conflicting opinions, no way to weight them, the same decision still in front of me. With AI as a thinking partner, the same hour looked different: I could ask for a sample of what other small business owners had actually said about rebuilding their sites, read through their real comments and complaints, and interrogate the patterns with the AI alongside me. By the end of an hour I had a clearer view than I'd had after weeks of solo reading.
With AI as a thinking partner, the same question went from "I don't have time" to "let me work through this for an hour." I could lay out my actual situation — small site, mostly informational, hosted somewhere I'd been on for years, performance issues — and have something working alongside me that could weigh trade-offs in real time. The constraint shifted from research time to decision-making time.
That's the actual unlock for small business owners. Not "AI builds your website" (it doesn't). Not "AI replaces your accountant" (it can't). It's: AI gives you back the hours you used to spend on the homework portion of any business decision.
The Conversations That Drove the Rebuild
I'll share three specific decisions I worked through with an AI tool, because the shape of the conversations matters more than the technical details.
Rebuild or patch? My first question wasn't "what should I build" — it was "should I do anything at all". I described the symptoms I was seeing. I asked for the strongest argument FOR patching and the strongest argument FOR rebuilding, and got both, fairly. That balance helped me see — for myself, after weighing it — that patching would buy me time but not a foundation. Rebuilding would give me something to grow into. For my situation, the answer was rebuild. AI didn't tell me to rebuild. AI showed me both sides clearly enough that I could see the answer myself.
What happens to all my existing content? The thing I was most worried about wasn't the technical work — it was losing my blog posts and the search rankings they'd accumulated over years. So I leaned into a habit I've found useful for any AI-assisted decision: I prompted the AI on what could go wrong, deliberately nitpicky, from every angle I could think of. "What if a redirect breaks? What if the URL format changes? What if Google reindexes before the redirects are in place?" Each answer either taught me something I hadn't considered or gave me one more box to tick before going live. By the end I had a verification checklist I'd built by interrogating the worst case, not by hoping for the best — and I'd learned more about how content migrations actually fail than I'd have picked up from any blog post on the subject.
What do I do if AI confidently recommends something wrong? This is the question that mattered most. I asked an AI tool directly: "How do I tell when you're confidently wrong versus when you're right?" The answer was: you can't tell from the AI alone — you have to check against reality. So I made a rule. Every recommendation got a verification step before being acted on. Twice that rule caught the AI giving me confident, wrong answers that would have wasted a weekend if I'd just followed them.
What I Had to Verify Myself
This is the part of the story that matters most for any business owner thinking about using AI for a real decision.
When the new site went live, I didn't trust that everything was correct just because it seemed to be working. I'd been building with AI long enough by then to know that "looks fine" and "is fine" are different things. So I spent the morning of the cutover running an audit of the live site. I checked image file sizes, how fast pages actually loaded, that every link still resolved correctly, that the contact form actually delivered, that the analytics correctly excluded my own office's network, and that the extra information Google uses to display the page in search results — the page title, the author, the date — was set correctly.
I found six things that were quietly broken. None of them would have shown up in a "does the homepage load?" check. Three of them, to give you a sense of the shape: an image that was loading at several times its proper file size, slowing the page significantly; a behind-the-scenes link tag that was telling Google this page was a duplicate of a different URL — small thing, big SEO consequence; and an analytics check that didn't quite work the way I'd thought. I'd deliberately set the site up so visits from my own devices wouldn't be counted — that was the whole point. But when I actually tested it during the audit, the setup wasn't catching everything. I worked through what was wrong with the AI, found the rule that needed adjusting, and tested again until it held properly. The other three bugs were similar in shape: small, easy to miss, real consequences if shipped.
The point is: I caught them because I was looking, not because the AI told me where to look. They were the bugs you get when you build anything substantial. The AI's job was to accelerate the build. My job was to verify it before the site went out into the world.
That's the deal with AI for any real business decision. Not a deal you should be scared of. Just a deal you should know the shape of going in.
What the New Site Does That the Old One Didn't
The outcomes are worth naming plainly:
- Pages load measurably faster on phones than they used to. The old site struggled to break 50 on Google's mobile performance test; the new one clears that comfortably.
- Internal traffic from staff devices no longer pollutes the analytics, so when I look at who's visiting, I'm looking at real visitors.
- The contact details on the page are accurate and current. Embarrassingly, the previous site had a partial old office address still showing in one place; this one doesn't.
- I can publish from anywhere, anytime — no plugin panels to wrestle, no waiting until I'm back at my desk. That's not a vanity feature; it's the difference between an idea making it onto the site this afternoon and the same idea getting forgotten by Monday.
- There's a proper AMA page with a QR code that anyone can scan to save my contact details directly to their phone — the kind of thing you wish every site had after meeting someone at a conference.
- When someone shares the URL on WhatsApp or LinkedIn, the preview card looks right — the title, the image, the description all show up correctly. That sounds minor; in practice it's the difference between a link people will click and one they won't.
None of this is fancy. None of this is novel. But it's all the kind of thing that quietly stops happening on small business websites that have been sitting unattended for a few years, and it's the kind of thing AI as a thinking partner makes easy to get right when you actually sit down and do the work.
The same approach applies to other decisions. I applied the same process when writing a Microsoft 365 buying guide for SMB owners on this site — same AI-assisted drafting, same human verification of every number before publishing.
Frequently Asked Questions
Can AI just decide for me?
No, and you shouldn't let it. AI can present trade-offs, surface options, and accelerate research, but the decision has to stay with you because you're the one who knows your business, your customers, and what risk you can afford. The trap is treating an AI recommendation as a finished answer instead of a starting point. Don't.
How long does a rebuild like this actually take?
For a small informational site rebuilt with AI assistance, mine took two days of focused work — but I had clear context about what the site needed to do, what content had to migrate, and what success looked like. Without that clarity going in, the same work could easily stretch over a couple of weekends. Most of the time isn't typing — it's deciding.
What about all my existing content?
Make a complete inventory before you change anything. Categorise everything — blog posts, service pages, landing pages, contact pages, any redirects you've accumulated. Plan the migration for each category specifically, and verify it on the new site before letting search engines see it. The work isn't hard. Forgetting a category is what hurts.
What does it cost to do something like this?
I haven't published a public costing because every business is different — what was right for me wouldn't be right for everyone. The honest answer is: the bigger cost is the time spent making decisions and verifying outputs, not the hosting bill or the tools. If you'd like to talk through what a sensible setup looks like for your specific situation, get in touch and we can have a real conversation.
What's the one thing AI can't do?
AI can't tell you when it's wrong. That sounds simple, but it's the entire game. AI tools are trained to sound confident — they will give you a wrong answer in exactly the same tone they'd give you a right one. The only way to handle that honestly is to verify outputs against reality before you act on them. Tests for the technical things, real-world checks for the factual things, common sense for the strategy ones. The verification is the human's job. The acceleration is what AI brings.
Want to See the Result, or Talk Through Your Own?
The new site is what you're looking at now. If you'd like to download the IT startup guide I wrote, scan the vCard QR to save my contact details, or register interest in upcoming AI workshops for small business owners, have a look at our AMA page.
If you're a small business owner thinking about your own website rebuild — or any major decision where AI could help — get in touch. The first conversation is just two people working through whether the project makes sense for your situation, with AI on the side accelerating the homework if it's useful. Reach me on 010 590 0090, email info@techcloud.co.za, or WhatsApp at wa.me/27105900090.