Talking to Your Website

Most personal websites die quietly. Not from neglect exactly — from friction.

You set one up, it looks great, and then updating it requires opening a code editor, remembering how the project is structured, pushing to git, waiting for a build. For a developer that’s manageable. For anyone else it’s a wall. Even for developers, that wall is just tall enough to make you think I’ll do it later — and later never comes.

I’ve been thinking about a different approach: what if updating your site was as easy as sending a text message?

The idea

Static sites are a great foundation. They’re fast, cheap to host, version-controlled, and don’t need a database. The problem is that the editing experience is frozen in the past — you’re expected to know Markdown, understand the folder structure, and feel comfortable in a terminal.

What if you layered an AI on top of that?

The concept is simple: a small local server with a chat interface. You open it in your browser and type what you want:

“Add a blog post about my trip to Austin”
“Add a Contact link to the navigation”
“Update my bio to mention that I now work with Python”

The AI reads your site’s current files, understands the structure, makes the edits, and commits them to git with a descriptive message. A CI/CD pipeline picks up the commit and deploys the updated site automatically. From your message to a live update: under two minutes.

Why this works

The key insight is that a static site’s content is just text files — Markdown, HTML, YAML config. AI is very good at reading and writing structured text. You don’t need a complex CMS with a database and an admin panel. You just need something that can translate plain English into the right file change.

Git gives you a full history of every edit. If the AI misunderstands what you wanted, one click reverts the change. Every update is traceable, reviewable, and reversible — better than most traditional CMS platforms.

What this unlocks

The interesting part isn’t the technology — it’s the behavior change.

When updating your site takes thirty seconds of typing instead of thirty minutes of setup, you actually do it. You add the blog post. You fix the outdated bio. You document the project you just finished. The site stays alive because the barrier is low enough that it never feels like a chore.

It also means you can update from anywhere. The chat interface runs in a browser. Your phone, a tablet, a borrowed laptop — if you can open a browser, you can update the site. The AI handles the file structure. You just describe what you want.

The stack

You don’t need much to build this:

• A static site generator (Jekyll, Hugo, Eleventy — anything that builds from text files)
• A free AI API (Google Gemini has a generous free tier, more than enough for personal use)
• A small Node.js server to bridge the chat UI and the AI
• Git for version control and change history
• A CI/CD service (GitHub Actions is free for public and private repos) to auto-deploy on push

The total infrastructure cost for a personal site running this way: close to zero.

The bigger idea

We spend a lot of time making technology easier to use for end users and very little time making it easier for the people who build and maintain their own corners of the web. The personal website shouldn’t be something you set up once and abandon. It should be a living document — easy enough to update that it actually gets updated.

AI isn’t replacing the developer here. It’s handling the part that was always just friction: knowing which file to edit, what the syntax looks like, where the content lives. You keep the creative control. You just stop fighting the interface.

That feels like the right division of labor.