Do You Need to Pay for a Custom Website Right Away? My Experiment with AI, GitHub, and Cloudflare

A lower-cost way to start a personal website, change it with AI, and keep the source code under your control. It's not one-click, and it isn't the right choice for everyone.

I've recently spent some time building and launching my own personal brand website, SuiK.

AI has changed the experience quite a bit. Instead of manually handling every layout adjustment and code change, I can describe the result I want, let AI help implement it, and then review what actually changed.

Once I connected ChatGPT, GitHub, and Cloudflare, I started thinking about a more interesting question:

If all you need is a place to showcase your work, explain your services, and help customers find you, do you really need to pay for a fully custom website from day one?

I'm not claiming to have invented a new website-building method. Several AI builders already offer more integrated experiences. I simply want to share a workflow I've used myself, including where it helps and where it becomes inconvenient.

Three reasons I think this is worth trying

1. Lower the upfront cost

A hairstylist, photographer, freelancer, or small studio might only need a portfolio, a short introduction, a services page, and contact or booking links.

For that kind of site, AI can help create a first version before you commit to a fully custom design and development project.

Lower cost doesn't mean zero cost. AI usage or subscriptions, a custom domain, advanced features, and future maintenance may still require money.

2. Make small changes without starting over

Imagine being able to say:

• "Use warmer colors on the homepage."
• "Add a gallery for short hairstyles."
• "Make the LINE booking link easier to find."

If your AI tool has the appropriate permission to edit the website repository, it can help make those changes and publish an update.

You still need to check the result. But you don't necessarily need to learn web development just to update your services or rearrange a portfolio.

3. Keep the things you've created

Facebook, Instagram, and YouTube are useful for discovery. But account access, platform rules, and reach are ultimately controlled by those platforms.

I like the idea of having a website that acts as a central home for my brand, while keeping its source code in GitHub and separately backing up my writing and original media.

One important distinction: GitHub does not automatically back up your social posts or recover your followers and comments. GitHub is also a third-party service, so important files deserve an additional local or independent backup.

The goal isn't to abandon social media. It's to use social platforms to reach people while maintaining a place you can continue to build and update yourself.

The three tools in my workflow

Tool	What it does
ChatGPT	Helps turn requirements into code and assists with changes and checks
GitHub	Keeps the full source code and version history
Cloudflare	Hosts the public site and can deploy updates from GitHub

I deliberately separate the AI, source-code repository, and hosting service. That means more setup, but it also makes it easier to understand where everything lives and what can be moved or replaced later.

The basic process

Here's the workflow I want to make accessible to people without a technical background. I still need to test how approachable it really is.

1. Prepare the accounts. Sign in to ChatGPT, GitHub, and Cloudflare. The AI tool you choose must actually be able to edit a code repository; access depends on the tool and plan.
2. Describe the website. Explain your brand, preferred look, pages, and what you want visitors to do. Reference images or websites are helpful.
3. Save the project to GitHub. Follow the official authorization flow and keep the site's source code in a repository you control. Don't paste passwords or private API keys into chat.
4. Connect Cloudflare. Import the GitHub repository using Cloudflare's deployment interface and complete the build and publishing steps. Some projects still require configuration; ask the AI to explain the exact fields if that happens.
5. Test the site and make one edit. Open the real URL, check the mobile layout, images, navigation, and booking links. Then ask the AI for a small change and verify that the repository and live site have actually updated.

You can start with a Cloudflare-provided subdomain such as *.workers.dev. If the website becomes part of your long-term brand, you can purchase a custom domain later rather than paying for one during your first experiment.

A sample prompt

I'm a hairstylist and I'd like a simple portfolio website. I want a warm, clean design that works well on phones. Please include an introduction, hairstyle gallery, services, Instagram link, and LINE booking link. Help me create the website, save the full source code to my GitHub repository, and deploy it to Cloudflare. Tell me which account permissions require my action, and clearly separate what you completed from what still needs testing.

That prompt is a starting point, not a promise that account authorization and deployment will happen without any user involvement.

When an all-in-one builder may be better

Tools such as Lovable, Bolt.new, and v0 already combine AI-assisted development with preview or publishing workflows. Some can also sync source code to GitHub.

If your priority is getting a website online as quickly as possible, with the fewest configuration screens, one of those products may be the better fit.

My approach makes a different trade-off:

• Potential advantages: a clear source-code location, greater freedom to change AI or hosting tools, and the possibility of lower recurring hosting costs.
• Costs: an extra round of account permissions, deployment setup, testing, and occasional technical troubleshooting.

It's a choice, not a universal recommendation.

Next: testing it with someone who isn't technical

I'm planning to try this workflow with a hairstylist who doesn't work in software development.

What interests me isn't just whether AI can generate a nice-looking page. I want to know where a real beginner gets stuck: GitHub access, deployment settings, or updating the site after it's live.

More importantly, would she actually want to put the result in her Instagram bio?

Building a site in 30 minutes is a challenge I'd like to test, not a result I've already verified.

If the process turns out to be too complicated, I should simplify it—or recommend a better tool—instead of expecting users to learn engineering jargon.

Final thought

The exciting part of AI-assisted development isn't only that engineers can write code faster.

It's that people who once saw a personal website as expensive or intimidating may now have a practical way to start.

Spend a little time, potentially save some money, and keep more freedom to update what you build. That's the idea I'm putting to the test.

SuiK. — From Ideas to Reality.