Paste a draft and get the exact tells: the reversal cliches, the em-dash cadence, the engagement bait, the flat rhythm. Deterministic rules, not a detector's guess.
0 words · checked locally in your browser
Paste a post on the left and the report appears here instantly.
The ruleset is the same deterministic lint that gates every draft inside Slingapult's composer, maintained like a security ruleset as the tells evolve. It looks at four things:
The "not X, it's Y" reversal, "isn't about X, it's about Y", "let that sink in", "I'm thrilled to announce", and the stock outreach openers.
"Delve", "game-changer", "tapestry", "a testament to", "unlock your potential", and the corporate filler verbs that mark text as generated.
Em-dash density far above how people write, emoji-bullet walls, hashtag walls, repost bait, and engagement-bait closers like a bare "Agree?".
Sentences all the same length, paragraphs all the same shape, and long posts with zero contractions. Humans vary; models average.
Originality.ai's study of 3,368 long posts from 99 influential LinkedIn profiles found 53.7% were likely AI-generated across 2025, and in trust-dependent categories human-written posts pulled roughly 40% more engagement. Readers have learned the tells. The risk to your reach is not a secret algorithm switch; it is sounding like the half of the feed everyone already scrolls past. We wrote up the full data in Does AI-written content hurt your LinkedIn reach?
One honest caveat: a lint can grade the writing, never the writer. A careful writer using AI as leverage will pass it. A human running on autopilot will fail it. That is exactly the standard your readers apply.
It runs a deterministic lint over your text: a maintained ruleset of known AI tells (signature constructions like the "not X, it's Y" reversal, vocabulary like "delve" and "game-changer", engagement-bait closers) plus computed checks for em-dash density, emoji-bullet walls, sentence-rhythm uniformity, and contraction rate. Same input, same result, every time. It is the exact gate that runs inside Slingapult's content composer.
No, and that is deliberate. AI detectors are probabilistic classifiers that guess at authorship and are often wrong in both directions. This is a lint: it flags specific, named constructions that read machine-made, whoever typed them. A human on autopilot can fail it. An AI draft that someone actually edited can pass it. It grades the writing, not the writer.
Originality.ai's analysis of 3,368 long posts from influential profiles found 53.7% were likely AI-generated across 2025, and in trust-dependent categories human-written posts pulled roughly 40% more engagement. Readers have learned the tells. When half the feed shares the same constructions, sounding like everyone else means being scrolled past.
No. The check runs entirely in your browser using the open ruleset on this page. Your text is never sent to a server, stored, or logged.
This gate runs inside Slingapult
Slingapult drafts posts in your own measured voice, runs every draft through this same lint until it reads human, and then captures everyone who engages as ranked, warm pipeline. Checking is free. Passing on autopilot is the product.
Get started