How This Blog Writes Itself (AI Cron + Gemini + GitHub API)

Full transparency: this post started as an AI draft. Here’s the pipeline that produced it — and why I still rewrote most of it myself.

The Pipeline

Three times a week (Monday, Wednesday, Friday at 10:00 UTC), a Vercel cron job fires a Python script hosted at /api/auto-blog. Here’s what it does:

1. Fetches trending topics from three RSS feeds: TechCrunch, WordPress News, and Shopify Blog
2. Sends the top story to Gemini API with a prompt that asks for a markdown blog post featuring the tools and services I offer
3. Pushes the file to blog/src/content/blog/ on GitHub using the GitHub API

The cron schedule lives in vercel.json, and the environment variables (GITHUB_TOKEN, GEMINI_API_KEY, and an optional CRON_SECRET) are stored in Vercel’s environment config.

The Honest Part

The AI generates decent first drafts. The structure is usually solid, the markdown formatting is clean, and it never forgets to include a CTA. But the voice, the specific details, the real numbers — that’s all me. I rewrite or heavily edit every post before publishing.

The drafts tend to be generic and too promotional. My job is to strip that out and replace it with honest experience: actual build times, real cost comparisons, genuine opinions.

Why I Built This

I’m a developer who builds Python automations for clients. This blog pipeline is a dogfood project — I use the same tools I recommend to clients. RSS scraping, LLM API integration, GitHub API automation, serverless cron scheduling. It’s a real-world showcase of what Python automation can do.

If you’ve got a content workflow, reporting pipeline, or data processing task that could benefit from this kind of automation, reach out. I build Python automations for a living.