Why I built my own comment system instead of reaching for Disqus

When I started this website a couple of years ago I built it from scratch primarily because I love playing with new technology and it was a great learning experience for me.
Today many people are choosing to build with AI rather than using off the shelf tools for simple things like static websites because the build vs. buy equation has radically shifted. If you can do it with AI in a few hours you might not want to use a third-party service any more.
I faced this question when thinking about how to build more community on this website. If you haven't noticed, I've been blogging a lot more lately. And while social networks like X and LinkedIn are where the conversation mostly takes place, I've been yearning for the good old days when blogs were king and we communicated in the comments.
In the past I would have just integrated Disqus on my site and called it done, but commenting is not the only community feature that I'm thinking about. I'd also like to add new post notifications and potentially a forum for discussions. I can't do this with the typical javascript-based commenting system.
Given my experiences with using AI, I wondered how hard it could be to just build it from scratch.
So I tried something different. I used my /new-issue skill to file a ticket describing what I wanted, then asked Claude Code to draft a plan. The plan covered the database schema, the auth flow, email notifications, and the component structure.
The first version was simple: a comment form, a database table, and an API route. I used Turso for the database because Claude told me it has a good free tier. It's SQLite over HTTP, so there's no server to manage. Within an hour I had comments saving and rendering on blog posts.
I needed a system for handling spam so I tossed around some ideas with Claude. We settled on Cloudflare Turnstile for bot protection, a moderation system for anonymous comments, and rolling our own magic link system for authentication with Resend.
After some back and forth and a little design feedback it was done. The whole project probably took 3 or 4 hours.
Now for most people, Disqus is probably still a good choice. It can be implemented with AI in a few minutes. But this experience proves that the bar is now much lower for many people to build systems shaped to their particular use cases.
While some could see the doom and gloom in this — the need for some software services is greatly diminished. But note that I replaced Disqus with not 1 but 3 services. The point is, you'll probably see a lot of services replaced by folks rolling their own with AI, but they will still need hosting, databases, email, and other services. The legos aren't disappearing. They are a different shape and more modular.
And for the end user this means no more waiting on roadmaps for features. Instead you'll turn more and more to AI to build custom software to meet your needs and desires.
So what do you think? Add a comment below to try it out!
Comments
You might also like…
Give your AI the full picture with a context repository
Poor AI output usually isn't a prompting problem, it's a context problem. A context repository gives your AI the business knowledge it needs to make good decisions. Here's how to build one.

Introducing Quiddity: generate essential skills for your dev workflow
Quiddity interviews you about your tools and process, then generates custom /new-issue, /next-task, and /approve skills tailored to how you actually work. One install, one setup command, and you're off.

Build a /new-issue skill and stop writing issues by hand
The final post in my series on skills for dev workflow. A /new-issue skill lets you describe a bug or feature in a sentence and get back a well-structured issue with wireframes and acceptance criteria — better than most people write by hand.

Extracting sub-skills from agent skills
As your agent skills grow, some parts want to be their own thing. Here's how I used a simple prompt to extract a standalone /approve skill from my /next-task skill — and why thinking of skills like functions leads to better workflows.
