Curro
AI content creation studio — from raw idea to published post, in your own voice. Voice-learned drafts, human approval gates, one-click publishing.
Curro started from my own frustration: generic AI writing tools flattened every founder voice into the same bland corporate tone, and my journal drafts took four hours each. I built Curro as an AI content studio that learns writing style from the user's own past work — vocabulary, sentence rhythm, formatting habits — then drafts articles, scripts, and social posts that sound like them on a good day, not like a chatbot. The pipeline runs idea → outline → draft → human approval gate → publish, with plagiarism and fact-flag checks before anything goes live. Under the hood it is a Laravel application with queued generation jobs, versioned drafts, and per-user style profiles stored as structured preferences rather than vague system prompts. I dogfood it daily: this portfolio's journal workflow — research briefs, audits, scheduled publishing — grew out of patterns I first prototyped in Curro. Measured on my own writing, a finished 1,300-word article went from roughly four hours to under 25 minutes of active effort, with quality gates catching thin or duplicated sections before publish. The roadmap I am working through now is multi-modal input (voice notes to drafts), team workspaces with shared voice profiles, and scheduled multi-platform distribution. If you publish expertise for a living, Curro is the compounding asset I wish I had bought years ago, so I built it instead. Stack: Laravel 13, queued jobs, Postgres with pgvector for style matching, Valkey caching. Timeline from first commit to paying users was eleven weeks of evenings alongside client work.
Visit project →
