The LinkedIn post generator that starts from what you already said
Paste a call, a case study, an article, or just talk for two minutes. You get three posts making three different arguments, written in your voice, with every number checked against your own words.
…so we pulled the onboarding data and 62% of the people who churned had never opened the reporting tab. Not once. And the ones who did open it in week one stuck around about eleven months on average.
Your churn problem probably isn’t a churn problem.
We pulled a client’s onboarding data last month. 62% of the people who left had never opened the reporting tab. Not once.
You have used one of these before. You typed a topic into a box, it gave you something confident and weightless, and you could feel that anyone in your industry could have posted it. So you edited it until it was yours, which took longer than writing it, and by week three you stopped opening the tab.
A blank box is the problem, not the feature
When a generator starts from a topic, it has to invent the substance. That is where the weightlessness comes from. There is nothing in the input that only you could have said.
PublishFlow starts one step earlier. It starts from the call you were on, the case study you ran, the thing you just said out loud. The specifics are already yours. The model’s job is to report what you observed, not to synthesise a take about your topic.
How it works
Bring something that happened
A call transcript, a case study, an article you wrote, a link, or two minutes of talking. Not a topic.
Voice notes record up to five minutes in the browser. Transcripts accept .vtt, .srt and .txt.
It picks the argument, then makes three of them
Your ideas are ranked against each other, best first, with a line on why. The one you pick becomes three posts committed to three different claims.
It checks its own work before you see it
The three first lines are tested against your readers. Banned phrasing is stripped. Any number that is not in your source gets flagged and fixed. When a hook is swapped, it tells you, and you can put the original back.
Hook rewritten after the reader test
The original opened on a definition. Your skeptical reader scrolled past before the number.
Three variations, three arguments
Not three openings on the same point. Each one commits to a claim the other two are forbidden from making.
From the same call
…so we pulled the onboarding data and 62% of the people who churned had never opened the reporting tab. Not once. And the ones who did open it in week one stuck around about eleven months on average.
Your churn problem probably isn't a churn problem.
We pulled a client's onboarding data last month. 62% of the people who left had never opened the reporting tab. Not once.
Breaks a belief the reader holds about their own numbers.
62% of the customers who left never opened the one tab that predicts retention.
The ones who did open it in week one stayed about eleven months. Onboarding never pointed them at it, so it may as well not have existed.
Explains the cause most people miss.
You'll lose another quarter of your accounts before the dashboard tells you why.
Churn reporting is a lagging report. By the time the number moves, the eleven-month customers you could have had are already gone.
Makes the cost of doing nothing concrete.
When the three posts were written independently they overlapped 47.5% word for word. Deciding the three arguments in one pass first brought that to 39.4%. That is the difference between a real choice and three versions of the same post.
Edit it without starting over
Change the hook, tighten a paragraph, shift the tone. Every edit works on the post you already have, so the parts you liked survive. Revert is always one tap away.
What it will not do
The limits, stated plainly, so you can decide before you sign up.
It will not post for you without asking.
Nothing is ever published without you approving it. Not on the trial, not on Pro.
It will not invent a number.
If a figure is not in your source or your own credentials, it gets caught and removed before you read it.
It will not make carousels or images.
Text posts only, on purpose.
It will not write about a topic you hand it.
It needs material. If you have nothing to paste, talk for two minutes instead.
Pro is 50 posts a month.
More than one a day. If you need more than that, this is probably the wrong tool.
Works with
Voice profiles
It reads your posts first, then writes like you on a good day.
Meeting notetaker
The bot sits in your calls, so the good line doesn't die there.
Voice notes
Say it out loud for two minutes. Get posts back.
The Monday digest
Monday morning, the week's posts are already drafted.
Questions
Is the LinkedIn post generator free?+
Everything is free for 14 days with no credit card. After that it is $29 a month for up to 50 posts, and your saved posts stay readable either way.
Does it sound like AI wrote it?+
It reads 3 to 10 of your own posts first and writes from your actual sentence style, hooks and vocabulary. A linter then strips the phrasing that gives AI away before you see the draft.
Can I edit the post it generates?+
Yes. Change the hook, tighten a section or shift the tone, and it edits the post you have rather than regenerating from scratch. Revert restores the original.
What can I feed it?+
A meeting transcript, a case study, a blog post or article, a URL, or a voice note up to five minutes.
Will it make up statistics?+
No. Numbers are extracted from your source before writing, and any figure in the draft that is not traceable to your material is flagged and removed.
Does it post to LinkedIn automatically?+
Only when you tell it to. Every post requires your explicit approval.
You already said something worth posting this week
Fourteen days free. No credit card.
Start freeJust want to try something quickly? Use the free post generator with no account.