← back to all work

Content EngineWIP

TypeScriptNextJSSQLiteLLMAutomationContent
Content Engine

Content Engine is a work-in-progress pipeline that turns Indonesian news into ready-to-post Instagram carousels, with a person approving every post.

 

A scout pulls news RSS feeds and Google Trends every 30 minutes, normalizes and deduplicates stories, and merges the same story from different outlets. An LLM analyst scores each story on relevance, emotion, novelty, timing and shareability and flags risky topics. Stories that pass go to a copywriter step that reads the full article and writes the slides, caption, hashtags and a fact-check list. A designer step renders the slides from templates, with optional AI cover art.

 

Everything is run from a dashboard: a desk of drafts waiting for review, a review page per draft with the slides, editable words and the fact-check list, plus pages for sources and topic presets, the niche brief, slide style (colours, typography and layout templates, cover-art styles) and AI spend. Auto-posting through the Instagram Graph API and analytics are next.

 

Built with TypeScript, Next.js, SQLite, Satori and resvg, with Anthropic or OpenRouter models.