Behind every website sits one of three main approaches — or a blend of all three. Understanding the difference means knowing what to choose, what it costs and what it implies. Here is the essence, from simplest to most advanced.

The three families in 30 seconds

• Static: ready-made files (HTML, CSS, images). Fast, safe, cheap.
• Dynamic: the server builds the page on every visit, often from a database. Flexible for managing lots of content.
• Web application: software running in the browser, connected to real-time data. Interactive and rich.

And increasingly, on top: AI and automations.

1. The static site

A static site is made of files generated once (at publish time), then served as-is to every visitor. No server-side computation: the server just delivers files.

• Strengths: very fast (nothing to compute), very safe (no database to hack), cheap to host.
• Limits: to change content you must rebuild the site. Less practical if non-technical people publish daily.
• Tools: generators like Astro, Hugo, Eleventy, Jekyll. This very site is static (Astro).
• Ideal for: showcase sites, blogs, documentation, portfolios.

2. The dynamic site (PHP, Apache, WordPress)

A dynamic site builds the page on the fly, on each visit. The server runs a program (often PHP), fetches data from a database (often MySQL), assembles the HTML and sends it.

The most common stack is LAMP: Linux + Apache (the web server) + MySQL (the database) + PHP (the language). WordPress, which powers a large share of the web, works this way.

• Strengths: an admin interface to publish without coding, customisable content, a huge ecosystem of plugins.
• Limits: more moving parts = more to secure and maintain (updates, backups), and potentially slower without optimisation (caching).
• Ideal for: content-heavy sites updated by non-technical people, shops, magazines.

3. The web application (Node.js, API, rich interfaces)

A web application is real software running largely in the browser (JavaScript), talking to a server through APIs. Think dashboard, booking tool, SaaS.

On the server side, Node.js (since 2009) lets you write that server in JavaScript — the same language as the browser. Interface frameworks like React, Vue or Svelte are often added.

• Strengths: deep interactivity, near-native app experience, real-time possible.
• Limits: more complex to build and host; overkill for a simple showcase site.
• Ideal for: interactive products, customer areas, business tools.

Where AI fits (Claude, Codex, automations)

Artificial intelligence replaces none of these families: it is added to all of them.

• Coding assistance: models like Claude (Anthropic) or Codex / Copilot (OpenAI/GitHub) write, explain and fix code. A developer goes faster; a beginner learns faster.
• In the product: chat assistants, smart search, content generation (descriptions, article drafts, translations).
• Automations: connecting tools together (via n8n, Make, Zapier) so AI writes relevant text, sorts emails, summarises a document — no fluff, on a trigger.

The right reflex: AI suggests, the human checks and decides. (See our AI pillar.)

The fusion: the modern web mixes everything

The line between these families is blurring. A high-performing site today is often hybrid:

• a very fast static base (the shell, delivered by a CDN);
• dynamic data fetched via API (a cart, reviews, stock);
• a headless CMS so non-technical people edit content without dictating the display tech;
• AI features grafted where they help (search, support, content).

This approach is often called Jamstack (JavaScript + API + Markup). The idea: take the best of each world.

Which one to choose?

Your need	The right approach
Present a business, a simple blog	Static (fast, safe, cheap)
Lots of content edited by non-technical people	Dynamic (WordPress) or headless
Interactive product, customer area, tool	Application (Node.js + framework)
Add value (content, support, code)	AI on top of the existing stack

The real skill is not choosing “the” technology, but combining these bricks for the goal — and not over-building.