AI Can Build Your Website. But Can You Actually Run It?

Entry 017

2025-07-02

Claude, ChatGPT, Cursor, Lovable, Replit, and other AI tools can generate functional websites quickly. For a simple landing page, prototype, or custom front end, they can reduce development time significantly.

The limitation is that generating a website is not the same as building the systems required to operate one. A business website may also need a CMS, eCommerce, payments, inventory, customer accounts, analytics, SEO controls, forms, hosting, databases, integrations, security, and ongoing maintenance.

These systems still need to come from somewhere.

1. You Still Need a CMS

A custom-coded website does not automatically include a content management system.

Without a CMS, changing basic content may require editing the code and deploying the website again.

This can include:

• Changing text

• Replacing images

• Adding pages

• Publishing blog posts

• Adding team members

• Updating locations

• Managing FAQs

• Editing SEO information

Platforms such as Squarespace, WordPress, Webflow, Wix, and Shopify include systems for managing content without editing code.

You can connect an AI-generated website to a headless CMS such as Sanity, Contentful, or Strapi.

That works, but the CMS still needs to be configured. Content types need to be created, fields need to be defined, the API needs to be connected, and the front end needs to display the content correctly.

AI can help build that integration. It does not remove the need for it.

2. A Product Page Is Not an eCommerce Store

AI can generate a product page with an image, price, variants, and an Add to Cart button.

A working eCommerce business needs much more.

It may need:

• Product management

• Inventory

• Variants

• Payments

• Checkout

• Taxes

• Shipping rates

• Discounts

• Orders

• Returns

• Customer accounts

• Transactional emails

• Fraud prevention

• Reviews

• Subscriptions

• Analytics

• Product feeds

• Fulfillment integrations

This is the infrastructure provided by platforms such as Shopify.

You can connect a custom website to Stripe, build a database, create an admin system, and connect separate services for shipping, taxes, email, and fulfillment.

For some projects, that makes sense.

For a standard online store, Shopify already provides most of it.

3. Someone Needs to Update the Website

This is one of the main limitations of a custom AI-generated website.

A business owner may need to:

• Change business hours

• Add an employee

• Update a service

• Publish a blog post

• Create a new product

• Change a price

• Add a promotion

• Upload new photography

• Create a landing page

A traditional website platform provides an interface for doing this.

A custom website may not.

If every change requires opening the codebase, prompting an AI tool, reviewing the change, and deploying it, the website is still dependent on a development workflow.

That may be acceptable for a developer.

It usually isn’t the best system for a client who needs to manage the website every week.

4. Custom Websites Require Infrastructure

A custom website needs somewhere to run.

A typical AI-generated website might use a combination of:

• Next.js or React for the website

• Vercel or Netlify for hosting

• GitHub for the codebase

• Sanity or Contentful for content

• Supabase or Firebase for a database

• Stripe for payments

• Resend or another provider for transactional email

• Cloudflare for DNS, security, or caching

These are all established tools.

The difference is that you are responsible for connecting and managing them.

With a platform like Squarespace or Shopify, much of this infrastructure is part of one system.

5. Forms Need a Backend

Generating a contact form is simple.

Processing the form requires additional infrastructure.

The website needs to determine:

• Where submissions are stored

• Who receives notifications

• How spam is filtered

• What happens after submission

• Whether submissions enter a CRM

• Whether automated emails are sent

• How customer data is secured

Website builders usually provide basic form handling or established integrations.

On a custom site, this needs to be configured separately.

The same applies to newsletter signups, applications, quote requests, booking forms, and other data collection.

6. SEO Requires More Than Page Copy

LLMs are useful for creating and editing website content.

SEO also involves technical implementation.

A production website should account for:

• Page titles

• Meta descriptions

• URL structure

• Canonical URLs

• XML sitemaps

• Robots directives

• Redirects

• Structured data

• Internal linking

• Image optimization

• Page performance

• Mobile performance

• Indexation

• Google Search Console

• Analytics

Established website platforms handle some of these items automatically and provide interfaces for managing others.

A custom website gives you more control, but that also means these features need to be implemented correctly.

7. Analytics Still Need to Be Set Up

A website being online does not mean it is being measured.

Depending on the business, a website may need:

• Google Analytics

• Google Tag Manager

• Meta Pixel

• Google Ads conversion tracking

• Form submission tracking

• Purchase tracking

• Add-to-cart tracking

• Checkout tracking

• Call tracking

• Cookie consent

• Other marketing pixels

For an eCommerce website, the tracking can become more complex.

AI can write the code required for these integrations, but someone still needs to define what should be tracked, implement it, test it, and verify that the data is accurate.

8. Integrations Need to Be Maintained

Most business websites connect to other software.

Examples include:

• HubSpot

• Klaviyo

• Mailchimp

• Salesforce

• Google Analytics

• Meta

• Google Merchant Center

• Stripe

• QuickBooks

• Booking software

• Inventory systems

• Shipping platforms

• CRM systems

A custom website can connect to these services through APIs.

Those integrations then become part of the website’s technical stack.

APIs change. Authentication can expire. Services update. Requirements change.

Someone needs to maintain those connections.

9. Custom Code Requires Maintenance

AI makes writing code faster. The code still needs to be maintained.

A modern website may depend on dozens or hundreds of software packages.

Over time:

• Dependencies are updated

• APIs change

• Packages are deprecated

• Security vulnerabilities are discovered

• Hosting requirements change

• Browser behavior changes

• Third-party services change

Established website platforms handle much of this work at the platform level.

With a custom website, more of that responsibility belongs to whoever owns the codebase.

10. Simple Websites Often Don’t Need a Custom Stack

A five-page service business website may need:

• Home

• About

• Services

• Blog

• Contact

• Forms

• Analytics

• Basic SEO

• An easy way to update content

You could build this with React, Next.js, Vercel, GitHub, a headless CMS, and several third-party services.

You could also build it in Squarespace, Webflow, Wix, or WordPress.

Both approaches can produce a good website.

The second approach may be easier for the business to manage.

Custom development should solve a requirement, not be the default.

When AI-Generated Websites Make Sense

AI development tools are useful for projects that require custom functionality.

Examples include:

• Prototypes

• MVPs

• Landing pages

• Calculators

• Interactive tools

• Internal applications

• Customer portals

• Dashboards

• SaaS products

• Custom applications

• Features that don’t fit an existing website builder

AI is also useful within traditional website projects.

It can help developers write custom components, CSS, JavaScript, integrations, scripts, and other functionality much faster.

The question is not whether AI should be used.

The question is what should be custom-built.

When Website Builders Make More Sense

Website builders provide infrastructure that most businesses need.

-Squarespace

Good for service businesses, restaurants, portfolios, local businesses, and straightforward marketing websites.

It includes hosting, content management, forms, blogging, analytics integrations, and basic SEO controls.

-Shopify

Good for businesses selling products online.

It includes product management, inventory, checkout, payments, orders, discounts, shipping, customer accounts, and a large app ecosystem.

-Webflow

Good for custom marketing websites that need more design control while maintaining a visual CMS.

-WordPress

Good for content-heavy websites and projects that need a flexible CMS and large plugin ecosystem.

-Wix

Good for small businesses that want content management and a broad set of website features within one platform.

The advantage of these platforms is not that they can generate a better-looking homepage.

They provide the systems required to manage the website after it is built.

AI Does Not Replace the Website Platform

AI changes how websites are built.

It can write code, generate layouts, create components, connect APIs, troubleshoot errors, and reduce development time.

What it does not automatically provide is the infrastructure behind a production website.

You still need to decide how content is managed, how products are sold, how payments are processed, where data is stored, how forms work, how analytics are tracked, how the site is hosted, and who maintains it.

Sometimes the right answer is a custom AI-assisted build.

Sometimes it’s Shopify.

Sometimes it’s Squarespace.

The platform should be based on what the business needs to operate the website.

The goal isn’t to generate a website as quickly as possible. The goal is to build a website the business can actually use.