Launching today
Articora
Write with AI in your own voice, and see where it drifts
2 followers
Write with AI in your own voice, and see where it drifts
2 followers
Articora learns your writing voice from a few of your own articles. When you write with AI in Articora, the AI writes with that voice profile, so drafts sound like you, not like everyone else. Every draft is checked against your voice paragraph by paragraph, with one line on why it drifts. Edits arrive as diffs you accept. Try the voice check free, no account: paste 2–4 of your articles and a draft.
Hi Product Hunt, Laszlo here, web dev.
What it is: a writing workspace where the AI writes with your voice. You train a voice profile from your own articles (upload, paste, or import from your blog), and every AI draft, edit and rewrite uses it. Then it measures itself: each draft gets a paragraph-by-paragraph read of where it still sounds like you and where it drifts. The AI proposes changes as diffs, nothing is rewritten silently.
The fastest way to see if it works for you is the free voice check, no login: paste 2–4 things you wrote plus a draft. Samples aren't stored, the result is a public link for 30 days. A sample from my own writing: https://articora.app/voice-check/5LIHBzCLw0xT9Cpy-ueRtOKh
Why I built it: I started using AI for drafts last year and my posts got... fine. Correct, readable, and somehow not mine. Every "humanizer" and "detector" I tried measures against a generic notion of AI text. I wanted the opposite: measure against ME. It's an estimate, not a verdict... on my own writing it separates my drafts from generic model output reliably, but I'd love to see it break on yours.
Honest limits: 2–3 short samples give a usable read, not a good one (the confidence label says so). It can't tell "you on a bad day" from "not you". Two voices in one set of samples get averaged silently, that's the next thing I'm working on.
It also keeps a record of what you typed vs. what a model touched. Free tier, no card; testers who give honest feedback get 3 months of Pro.
Would genuinely like to hear where the score is wrong for you. That's the useful feedback.