We publish a lot of content. Here's how the pipeline works — from idea to published post.
THE STACK
- Database: MySQL via Prisma ORM —
blog_posts,blog_categories,blog_tagstables - Content: HTML stored as LongText — full control over markup
- Images: A2E.ai API for cover generation, Sharp for responsive WebP thumbnails
- Typography: Editorial passes on long-form posts — proper heading hierarchy, line heights, and reading rhythm
- Delivery: ISR (Incremental Static Regeneration) on public blog routes
THE PROCESS
- Seed script — TypeScript script creates post with Prisma, generates unique ID and slug
- Cover image — A2E text-to-image generates a dark, themed cover matching the post subject
- Typography pass — Headings get font-black uppercase treatment, body gets clean Geist Sans
- WebP generation — Sharp converts covers to responsive WebP thumbnails
- Publish — Status set to PUBLISHED, ISR caches the page, RSS feed updates
WHAT WE'VE LEARNED
Seed scripts beat manual CMS entry for batch content. The editorial typography pass is critical — raw HTML from AI tools needs consistent formatting. And WebP thumbnails cut page weight by 60% versus PNG covers.



