Someone on your team opened ChatGPT, described the business, and a few hours later there was a website. It has your colors. It has your services. The copy sounds close enough. It genuinely looks like the thing you've been meaning to build for two years. So what happens now?

I get this call more and more often, and I want to answer it honestly, including the parts that aren't great news. Some of you can publish what you have and get back to work. Some of you are further from the finish line than the screen suggests. The difference between those two groups has almost nothing to do with how the site looks.

What you actually built

First, credit where it's due. You made a decision and produced something real, quickly, without a six-week discovery process. That's not nothing. I've watched businesses spend two months and real money arguing about wireframes and end up with less clarity than you got in an afternoon.

Here's the useful way to think about what's now sitting on your screen: you have an unusually good brief. You've answered the hardest question in any website project, which is what do you actually want this thing to be. You can point at it. Your team can react to it. Nobody has to imagine anything.

What you don't necessarily have is a finished website, and the gap between those two is where people get surprised. An architect's rendering is genuinely valuable. It shows the light, the proportions, the feel of the rooms. It also doesn't include the gas line, the load calculations, or the permit. Nobody thinks the rendering is the house. But when the rendering is clickable and lives at a web address, it's very easy to think it's the house.

One more thing worth knowing: not all AI-built sites are the same underneath. A site described in a chat window and a site built with an actual coding tool can look identical in the browser and be wildly different in the code. That difference doesn't show up until someone needs to change something.

The visible half and the invisible half

A business website is really two products stacked on top of each other. There's the part your visitors see, and there's the part that makes it an asset instead of a picture.

AI has gotten genuinely good at the first part. Layout, hierarchy, color, reasonable copy, a contact form that appears on the page. Two years ago that took a designer a week. That's a real change and I'm not going to pretend otherwise.

The second part is where things thin out. Not because the tools are bad, but because most of that work isn't visible in the output, so there's nothing prompting anyone to ask for it. It's plumbing. Nobody photographs the plumbing.

Nine things to check before you call it live

Run through these yourself. It costs you an hour and it will tell you far more about where you stand than any opinion I could offer from the outside.

1. Does the contact form actually deliver? Fill it out with a real address. Then check your spam folder. A form that renders is not a form that works, and this is the single most common failure I find.
2. Is it on your own domain, with a valid certificate? Not a preview link, not a subdomain of the tool you built it in. Yours.
3. Does Google know it exists? Search for site:yourdomain.com. If nothing comes back, you're invisible, and that usually stays true until someone fixes it deliberately.
4. Does it work on a real phone, on cell data, away from your wifi? Not the browser's phone preview. An actual phone in a parking lot.
5. Can someone navigate it with a keyboard or a screen reader? For a business in the United States, accessibility isn't a nice-to-have. It's exposure.
6. Where does customer data go when someone submits it? If you're in healthcare, law, or financial services, this question has legal weight and a wrong answer is expensive.
7. Does it connect to anything you already use? Your CRM, your calendar, your payment processor, your email list. If a lead lands in an inbox and stops there, you've automated nothing.
8. Can you change it without fear? Try editing something small. If you're nervous about touching it, that's information.
9. What happens when it breaks on a Saturday? Who do you call, and can they get at the code?

If you sailed through all nine, you're in genuinely good shape. Publish it. I mean that.

Why the last 10 percent is usually most of the work

Here's the conversation I'd rather have on a blog post than on a call with you, because on a call it's uncomfortable for both of us.

People arrive believing they're 90 percent done. Fairly often they're closer to 20 percent. Not because anyone did anything wrong, but because the visible layer, the part that got built, is roughly a third of a working business website. The rest is integration, data handling, performance, security, indexing, and making the thing maintainable by a human being next year. That work doesn't look like anything. It just costs hours.

If you get a quote that's dramatically higher than you expected, this is almost always why. It's not that your site is bad. It's that the part you can see was never the expensive part.

I'd rather tell you that now, for free, than have you feel ambushed by a number later. And I'll say the other half of it too: sometimes the honest answer is that rebuilding on top of your vision costs less than repairing what's underneath it. That sounds like a wasteful thing for someone in my position to say, but patching code that nobody can explain has a way of costing more over eighteen months than starting clean did on day one.

When you're genuinely fine

I'd rather you not spend money you don't need to spend, so let me be specific about this.

If your site is a personal project, a portfolio, a landing page to see whether an idea has legs, or a simple presence for a business where the real work happens by phone and referral, and no money moves through it, and you're not collecting sensitive information, and you can edit it yourself when something needs to change, then you're fine. Ship it. Come back when the business has grown into needing more.

That's a real answer, and it applies to a lot of people reading this. Not every website needs a professional, and I'm not interested in convincing you otherwise.

When it's time to bring someone in

The picture changes when any of these are true:

• Money moves through the site, or it's how you book paying work
• Customer information lands in it, particularly health, legal, or financial details
• It's the main way new customers find you, so being invisible in search has a measurable cost
• You're in a regulated industry where accessibility or privacy compliance is enforceable
• It needs to talk to the systems you already run on, and right now it doesn't
• You can't change it without worrying you'll break it
• Your own time has become the bottleneck, and every hour you spend fighting it is an hour not spent on the business

That last one is the one people underrate. I've watched owners spend three weekends trying to get a form to deliver reliably. Their time was worth more than the fix.

What I'd do in your position

Keep what you built. Seriously, don't throw it away, and don't let anyone tell you it was wasted effort. It's the clearest brief you'll ever hand a developer, and it will save you real money in a way that's easy to overlook.

Then get an honest assessment before you get attached to a number in your head. Ask someone to walk the nine points above with you and tell you which ones are actually a problem for your situation. A good answer will be specific and some of it will be "this one doesn't matter for you."

And ask about the foundation, not the finish. The question isn't "can you make this look better." It's "what will this cost me to own for the next three years." If the answer to that is uncomfortable, better to know now, while your only investment is an afternoon of prompting.

If you want a second set of eyes on what you've built, that's a conversation I'm happy to have, including the version where I tell you that you don't need me yet. You can see how I approach web design and development, or how the same thinking applies to bringing AI into a business in a way that survives contact with reality.