Newsletter operators: what would AI need to get right before you trust it anywhere near your voice?

Hey 👋

Cagri here, co-founder of HeyNews. We go live in about 10 hours (May 12, 12:01 AM PT), and I wanted to start a real conversation in here first instead of saving everything for the listing.

Quick context on what we've been building:

HeyNews is a newsletter production platform that learns your editorial voice from your past issues.

You connect your archive (beehiiv, Kit, or any newsletter with a public archive like Substack, Ghost, Mailchimp, or Medium), and we train a voice profile based on how you actually write: tone, sentence rhythm, and the phrases you reach for.

It pulls stories from sources you trust (RSS, blogs, social, Reddit, saved articles via a Chrome extension) and gives you a publish-ready draft.

You review and send. We never auto-publish.

We used it ourselves for a year before deciding to launch. Nearly 600 issues across 10+ formats. That's the only reason we're comfortable opening it up now.

The thing I'd love to hear from people here before tomorrow morning:

What would AI have to get right before you'd trust it anywhere near your newsletter voice?

Not what's on our roadmap. What's the hard line for you? The thing where, if a tool got it wrong even once, the trust is gone for good.

I'll read every reply. @eren_daskesen (my co-founder) is here, too. Happy to answer anything about HeyNews, but I'd rather start with your answers.

Cagri