One article becomes five ready-to-post versions
A LinkedIn post, an X thread, a newsletter section, the search listing and three pull quotes — all written from the same article in one pass, and all agreeing on every number in it.
Tested 26 Aug 2026 on one 600-word article: 5 of 5 versions produced, every figure traced back to the source.
How it works
Five steps, left to right. Click the picture to see any step larger.
Your article goes in
One long piece — a blog post, a report, a case write-up.
- Title
- Why Most Onboarding Emails Fail
The facts are pulled out once
The AI (Claude) reads it a single time and lists what it says: the subject, the audience, the claims, the numbers.
- Main claim
- Trial-to-paid moved from 11% to 19%
- Strongest number
- 1,200 sequences analysed
Five writers work at the same time
One per place you publish, each with its own rules on length, opening line and tone.
- 120–200 words, no hashtags
- X thread
- 5–7 posts, each under 260 characters
The five are brought back together
All five arrive, in order. If one fails, the other four still ship and the pack names the one that's missing.
One pack, ready to paste
Each version under its own heading, so you copy the LinkedIn one into LinkedIn and you're done.
- Pack header
- Channels produced: 5 of 5
- Missing
- none
How it's built
Three decisions behind it, and the first one is the whole idea.
The article is read once, not five times
The simple build gives the whole article to five writers and lets each read it for itself. Five readers drift: one rounds 2.1% to 2%, another remembers the sample as "over a thousand", and your LinkedIn post and your newsletter now disagree in public. Here the facts are pulled out once into one shared list, and all five writers work from that list only. They cannot contradict each other on a number, because none of them is reading the article.
Every writer gets its own brief
A LinkedIn post and an X thread are not the same text at different lengths. Each writer has its own rules — opening line, length, whether hashtags are allowed, how to close — so the output is five genuinely different pieces rather than one post cut into five shapes.
The step that collects the five says five, in writing
The step that brings the five versions back together expects two by default. Left at the default it would quietly keep two and drop three — no error, no warning, just a pack that looks finished and is missing most of itself. The build states the number five explicitly, and the test run counted five in and five out to prove it.
What was verified
Run end to end on 26 August 2026 on one sample article of about 600 words carrying a dozen specific figures. All five versions were produced in a single run, and every number in every version was checked back against the article. None was invented, rounded or shifted.
The items were counted rather than assumed: five versions in, five out of the step that collects them, one pack. 14 steps, 26 of 26 connections wired, no errors.
Not yet verified: it reads one built-in sample article, so it hasn't been run on a live feed or a batch at once. The failure path — one writer dying while the other four ship — is built and wired, but was not triggered on purpose, so it is untested in practice.
From the test run
"Sending fewer, shorter emails outperformed longer sequences in every single test." — one of the three pull quotes, each written to stand alone on a graphic.
The same 11% to 19% figure appears in the LinkedIn post, the thread and the newsletter, written the same way in all three. That is the shared fact list doing its job.
