Writers are right to ask hard questions about AI. Here are our answers - specific, checkable, and built into how the product works.
Quill never writes your words. Not a chapter, not a scene, not a line of dialogue, not a logline, not a query letter, not a synopsis. Nothing Quill produces is designed to be pasted into your draft, and nothing it says is presented as your writing.
What Quill gives you is what a good editor gives you: notes, questions, structural observations, and directions. A coverage report will tell you a logline is muddy and point at why - it will not hand you a new one. That limit isn't a style choice in our prompts; it's enforced in the product, and we test for it.
Early on we considered a carve-out: let Quill draft query letters and synopses - the sales paperwork around a story, not the story itself. It would have been easy to build and easy to justify.
We said no. One exception makes every promise on this page negotiable, so the rule stands without exceptions: guidance everywhere, ghostwriting nowhere. Quill will coach your query letter line by line. It will not write it.
We want to be precise, because a vague “guidance only” claim wouldn't survive a look at the product. Alongside the coaching, One More Draft can generate certain things: character portraits, storyboard panels, world maps, and audio table reads.
The line we hold: those are renderings of your work, never the work itself. Each one is made only when you explicitly ask, only from your own material, and is always labeled for what it is. Your words stay yours to write; the pictures and voices exist to help you see and hear what you wrote. If you never touch those features, nothing about them touches you - every AI feature here is opt-in, per use.
We never train AI models on your writing - we don't train models at all. When you ask for feedback, the relevant material is sent to our AI provider to produce that one answer; Quill's text features run on Anthropic's commercial API, which does not train on customer inputs by policy. We keep prompts only as long as needed for cost tracking, debugging, and abuse prevention - never for training. The same commitment applies to the media features above.
Your work is also never used to shape anyone else's results. Feedback for you is generated from your project alone - there is no shared pool of members' writing behind the scenes. And you own everything you write here, full stop (see the Terms and Privacy Policy).
The most important feedback on this site comes from people. The Workshop is peer review - writers reading writers, with reputation earned by giving real notes. And the Marketplace exists to connect you with paid human professionals - script consultants, editors, coaches - because we think the answer to “will AI replace readers?” should be a product that pays human readers.
You'll notice the site says “Quill” where you might expect “AI.” That's not concealment - this page is in the footer of every page on the site. It's a design commitment: Quill is a named tool with a defined job (coach, never ghostwriter), and giving it a name keeps us honest about that job. If Quill ever did start writing for you, the name would stop being true.
Some writers want no AI anywhere near their process. We think that's a legitimate position, and this site works fully without touching a single AI feature - the frameworks, the editors, the Workshop, and the community don't require it. If you think we've drawn a line in the wrong place, we genuinely want to hear it: tell us.
