How these pages are written — and how to get one corrected.

Who writes, who checks

The voice on these pages is Petalie's — the AI social media teammate that watches TikTok, Instagram and Facebook videos. She writes in the first person because that is who answers you inside the product. Every page carries a byline that says so. confirmed

The research behind the /learn pages — what each platform has actually published about its feed — is Jean-Paul Azzi's, the founder; it is the plain-language version of a formal study that is becoming a book. He reviews every page before it goes up and every change after. An AI wrote the sentences; a person is accountable for them. confirmed

What counts as a source

For anything about how a platform's feed works: the platform's own engineering paper, its official page or newsroom post, or its open-source code. Nothing else counts as evidence — not a marketing blog, not a creator's theory, not a tool vendor's guide, not a screenshot. Google's own guidance on helpful content asks for the same things: who wrote it, how, and where the facts come from. confirmed · 1

• confirmed — the source says it, and the tag links to that source.
• inferred — our deduction from what is confirmed; the reasoning is in the sentence.
• speculated — a guess, and the page says so in as many words.

Each source is dated and labelled by kind (paper, official page, open-source code) and status (production, research, official statement). An old source stays old on the page: a 2016 paper is cited as 2016, with a note that systems change. confirmed

What we say about Petalie

Only what shipped. The free allowance on a page (how many videos a guest read covers, how many messages a session holds, what the 7-day Creator week is) is read from the product's own configuration and stated in numbers, and the page's date is the day those numbers were checked. What she cannot see is written down next to what she can: the retention graph, who the platform showed a video to, private accounts. confirmed

When we compare Petalie with another tool — ChatGPT, a hook analyzer, a follower checker — we say only what we tested ourselves, with the date of the test, in the tool's own current version. We do not describe competitors' features from their marketing. confirmed

No page promises reach, virality or growth. Her feel for how a video will do comes from a model trained on real videos and their outcomes; it is described as a feel, never as a prediction with a guarantee. About one video in four gets 'hard to call'. confirmed

Dates, updates and corrections

Every page shows the day it was last changed, and the change is real: a page that was only rebuilt without a change in content keeps its date. A correction is made on the page, with a line saying what changed and when, so a reader who saw the old version can find the difference. confirmed

To report a mistake — a wrong number, a claim you can show is out of date, a source we misread — write to [email protected] with the page's address and, if you have it, the source. We answer, and we fix the page, usually within a few days. confirmed

AI, crawlers and your data

These pages are written by an AI and reviewed by a person, and say so. Search engines and answer engines are welcome on every public page — the same robots.txt group applies to all of them, as the standard requires — and are kept out of the app itself. confirmed · 2

A guest read on a public account uses only what that account shows to anyone. Nothing from inside your Petalie room is ever used on a public page. The privacy policy has the rest. confirmed