
Writing You can start this yourself today.
Your voice, measured. Your claims, checked.
Will it sound like you? Will it say something untrue? Jinn listens first, scores every post against your voice and keeps a receipt for each step.
At a Glance
Read your brand, draft, check, publish.
- What it listens to
Four sources: RSS, Reddit, Hacker News and X. A daily run scores what it finds against your brand before anything is drafted. More sources are coming.
- How it writes
From your voice record, read in two passes: one for what you write about, one for how you write it. The first draft starts in your voice instead of repeating your topics.
- What it checks
Every post passes your voice checks before it publishes. Longer pieces also clear a structure check and a fact check that reads every source you cite.
- What it keeps
A receipt for every draft: where it came from, what you approved and what went out, kept in a chain that shows if anyone changes it later.
Every Writing draft follows one loop: your record, the draft, the check, then your yes.
Example brands, real Jinn output
Test line added to show the check
The note stays with the draft until you settle it.
Jinn checks a claim against your own record. It does not decide whether a claim is legally allowed; your reviewer still does.
Your yes, after the check
The draft reaches you with the marked line and its note. Fix it, keep it or send it back.
Example brands, real Jinn output
Listening
It listens before it writes.
- Four sources, named
RSS, Reddit, Hacker News and X. Those four exactly, with more coming, so you always know where a signal came from.
- Scored against you
Each signal is scored for how well it fits your brand, and only what clears the bar becomes a draft. The market is loud; the queue stays on topic.
- Standing, daily
This runs every day, not when you remember to press a button. You wake up to the opening worth taking, already read for you overnight.
Voice
A brand voice is more than just vibes.
Your team worked hard to design a brand voice and tone that suits your brand. Jinn helps you scale your branded content, while staying on-brand.
Generic AI reads your old posts and repeats your subjects. Jinn works in two passes. The first pulls out what your samples are about (your topics, facts and stories) and sets them aside. The second studies how you write, with that topic list held out, so it learns your style instead of echoing your last campaign.
Nothing in your voice record is invented. A fact earns a place only if it appears word for word in your own samples; anything without its quote is dropped before the record is built. Your voice is learned from you, not made up to sound plausible.
And it is tested. The same voice on new topics should read as one person, and two voices on the same topic should read as two. That check is a real test in the code, not a claim on a slide.
The checks
The checks that matter most run as code.
Before anything reaches your queue it clears a set of checks. The most important ones are rules that run the same way every time and can’t be talked around by a well-phrased draft.
Voice, on every post
Every post, short or long, is scored against your voice record before it can publish. A post that misses your voice waits for a fix.
Structure, on longer pieces
Longer pieces also clear a structure score for the shape AI answer tools quote: sections that open on their own, a takeaways block, a real list of questions. No AI grades it. Short posts aren’t held to this bar.
Sources, actually opened
When a longer piece cites a number, Jinn opens that source behind a safety guard and checks the number is really on the page. A citation the source can’t back is flagged on its own, without failing the rest.
Compliance holds when unsure
Where you switch compliance on, three reviewers read each draft: regulatory, brand and voice. The regulatory reviewer checks it against the FDA and FTC rules Jinn tracks. A check that can’t finish holds the draft; it never counts as a pass.
Which is which
What runs in code, and what a model reads.
Where you switch compliance on, the checks in code run first. Then three narrow model reads add what they find. A model can add a finding and never remove one.
| Check | What it looks for | If it fails |
|---|---|---|
| Banned words, in code | The words your brand never uses, matched exactly | A must-fix finding. A person fixes it, or overrides it with a written reason. |
| Category rules and your disclaimers, in code | Your category's rule packs, the disclaimers you require and the claims you marked risky | A finding at the rule’s own severity, for a person to decide. |
| Claims, read by a model | Each claim, against the FDA and FTC rules Jinn tracks | A finding, added beside the ones the code found. |
| Your rules, read by a model | The brand rules you wrote | A finding, added beside the ones the code found. |
| Your voice, read by a model | Whether the draft sounds like your voice record | A finding, added beside the ones the code found. |
| A check that can’t finish | Any of the above | The draft holds. It never counts as a pass. |
Receipts
Every draft keeps its receipts.
Every draft is labeled with where it came from the moment it is made: written by hand, promoted from an idea, pulled from listening, split from a longer source, or drafted overnight. The label is set when the draft is created, not guessed later.
After that, every change is added to a trail that is never edited, and each approval or publish saves a copy of exactly what was on screen, including the compliance verdict showing when you pressed the button. Months later you can still answer “what did we approve, and what went out?”
Each entry is sealed to the one before it, so changing an old entry breaks the seal and the check catches it. It is the paper trail a regulated brand needs, and nobody, including Jinn, can quietly rewrite it.
Honest machinery
What Jinn won’t do with your name on it.
No posing as your customers
Jinn keeps a written list of what it won’t do: no fake accounts, no posing as customers, no engagement it wouldn’t disclose.
No facts the source doesn’t hold
Hand Jinn a long post or a YouTube talk and it pulls the distinct angles worth writing. A check then drops any number or quote the source doesn’t contain, and a source too thin to work from is turned down.
No record of its own
Every draft starts from your brand record: 700+ signals drawn from your own site and material. Jinn reads that record. It doesn’t make one up.

Nothing ships without your yes.
Pockets, Head of Approvals & Keeper of the PenQuestions
The questions people ask.
Does Writing post anything without my approval?
No. Writing drafts and queues, and a post goes out after you approve it. Every post clears your voice checks first, plus the compliance check wherever you switch it on.
Which platforms does it publish to?
X, LinkedIn, Bluesky and Facebook. Performance reports cover X only today, and this page won’t pretend otherwise.
What gets checked before something goes out?
Your voice, on every post. On longer pieces, the structure and every cited number too. With compliance on, the three reviewers as well.
Where do the drafts come from?
From what Jinn hears each day on RSS, Reddit, Hacker News and X, and from what you hand it, like a post or a YouTube link. Each draft carries a label saying which.

Writing: want to see it run on your brand? Let's talk.
- Tell us about your brand.
- A person on our team reads it.
- We read every application and reply within two business days.
Already use Claude or ChatGPT? Connect your AI, free