AI, Agents, and 'Vibe Coding' — Building Gatherly in Days, Not Weeks

Lately there’s been a lot of talk about “Vibe Coding” — the idea that you can build complex software simply by explaining what you want, iterating conversationally with an AI, and letting it handle the scaffolding.

I wasn’t sure what to make of it… until I built Gatherly.

What Is Gatherly?

Gatherly is a lightweight, self-destructing event page service — think no registration, privacy-first, shareable links that vanish after the event.

It’s inspired by the excellent Gath project but without the Fediverse/ActivityPub layer I didn’t need. It’s simple, minimal, and self-hostable — built with Go, HTMX, and BeerCSS.

The story began when a friend in the Yarn.social community suggested we restart our monthly online calls. Previously, managing RSVPs was manual — posts, replies, and mental checklists. I wanted something frictionless.

When I couldn’t find the right fit, I decided to build my own.

Building Gatherly with an AI Agent

I opened VS Code, started a new project, and used OpenAI’s CodeX agent.
I described the idea, the user journeys, the tech stack, and UI style — and then let the AI do the heavy lifting.

Over the course of just a few days — with careful prompting, context, and iteration — I had a working prototype, CI/CD pipelines, tests, documentation, and a production deployment.

Normally, this would’ve been a weekend-long solo grind. Instead, I spent less time typing and more time engineering — thinking, refining, testing.

So… Is This “Vibe Coding”?

I’m still deciding.

Is “Vibe Coding” simply good engineering amplified by AI assistants, or is it something new entirely?
All I can say is that the result — Gatherly — is production-ready, privacy-respecting, and polished, and it only took days, not weeks.

💡 Built with Go, HTMX, BeerCSS, and a bit of AI magic.

Closing Thoughts

What do you think — is Vibe Coding the next buzzword, or a genuine evolution in how we build software?