Why this matters more than it did a month ago
On 30 July LinkedIn added a "seems like AI slop" button to posts. People pressed it more than a million times in the first two weeks, which Fortune reported as proof of how tired members are of machine-written text. LinkedIn has also said posts it classes as slop now get about 40 percent fewer views.
Messages do not have the button, but the people reading them have the same reflex. A connection note that opens with "I hope this message finds you well" gets the same half-second judgement: this was not written for me. Once someone thinks that, the request is ignored, however good the offer behind it.
This tool reads your message the way a busy stranger would. It points at the exact phrases that give it away, tells you why each one reads as automatic, and rewrites the whole thing in plain words that still sound like you. It is free, it needs no account, and it takes about ten seconds.
The tells that make a LinkedIn message sound like AI
No single word proves a message came from a chatbot. What gives it away is a pile-up: an opener anyone could have written, praise that fits every profile, three neat benefits in a row, and not one detail that could only be about the person reading it.
We score the whole message from 0 to 100 on how it reads, then flag the phrases doing the most damage. Specific, checkable details pull the score down more than anything else. A message that could go to a thousand people unchanged pushes it up, even when no single phrase is bad.
The rewrite keeps your goal, your facts and any merge fields like {{first_name}}. It never invents a compliment or a detail about the other person. If your draft has nothing specific in it, the rewrite leaves a marked gap for you to fill instead of making one up.
Generic openers
"I hope this finds you well", "I came across your profile", "I wanted to reach out".
Stock phrases and buzzwords
"Explore synergies", "innovative solution", "unlock", "elevate", "game-changer".
Long dashes
People rarely type an em dash in a DM. AI drafts are full of them.
Lists of three
"Faster, smarter and more efficient", especially twice in one message.
Forced enthusiasm
"I'm thrilled", "truly impressive", praise nobody has earned yet.
Too polished
Every sentence the same length, a neat summary line, "not only X but also Y".
Three LinkedIn messages, rewritten to sound human
These are the kind of drafts people paste in, and the shape of what comes back. The rewrites are shorter, name one real reason for writing, and ask for one small thing.
The real details in the examples (the post about hiring, the lost customer) have to come from you. When your draft has none, the tool leaves a marked gap rather than inventing one.
| Kind | Before | After |
|---|---|---|
| Connection note | Hi Sam, I hope this message finds you well. I came across your impressive profile and would love to connect and explore potential synergies. | Hi Sam, saw your post on hiring your first sales rep. We are about to do the same. Happy to connect. |
| Follow-up | Just circling back on my previous message. I'd love to find a time to connect and explore how we can drive results together. | No worries if the timing is off. If hiring a rep is still on your list for Q4, I can send the checklist we used. |
| Post opener | I'm thrilled to share three powerful lessons about leadership, resilience, and growth. | We lost our biggest customer in March. Here is what we changed. |
LinkedIn character limits, and the length that gets replies
The tool counts your characters as you type and checks the rewrite against the limit for what you are sending. LinkedIn's own help page puts the connection note at 200 characters, and says free members can add a note to three requests a month. Some guides still quote 300 for Premium, so we write every rewrite to fit 200.
The limit is the ceiling, not the goal. The messages that get answered are usually far shorter than LinkedIn allows.
| What you send | LinkedIn limit | What we aim for |
|---|---|---|
| Connection note | 200 characters (300 on some paid accounts) | Under 200, one reason, no pitch |
| Message to a connection | 8,000 characters | Under 500 |
| InMail | 200 subject, 1,900 body | Under 600 |
| Post | 3,000 characters, about 210 before "see more" | The hook inside the first 210 |
How the LinkedIn AI detector works
- 1
Pick what you are sending
A connection note, a message, an InMail or a post. Each has its own limit and its own idea of too long.
- 2
Paste your draft
Straight from LinkedIn, your notes app or the tool that wrote it. Merge fields are fine.
- 3
Read the flags
Each robotic phrase is highlighted in place, numbered, with one line on why it gives you away.
- 4
Take the rewrite
Copy it as it is, or use it as a starting point. Its length is checked against LinkedIn's limit.
- 5
Check the reply hint
A rough read on whether a stranger would answer the original. It is a hint about how it reads, not a forecast.
- 6
Keep the full pack
Leave your email for two more versions, a follow-up message, and a fix for every flagged line.
What this tool is, and what it is not
We would rather tell you the limits up front than have you find them later.
- It is a style check, not a forensic AI detector. It judges how your text reads, not who or what wrote it. No tool can prove authorship from a few lines, and anyone who says theirs can is overselling.
- A low score does not mean LinkedIn will treat your message well, and a high score does not mean you used AI. Plenty of people write stiffly on their own.
- The reply hint is an opinion about the wording. Who you write to and why matters far more than any sentence.
- We do not store your message. We keep the score and a one-line summary so your share link works, and the rewrite for 7 days so we can email it if you ask.
- It reads English best. Other languages work, with less nuance.
- Ten checks an hour per person keeps it free for everyone.
Frequently asked questions
Straight answers on how it works, how it keeps your account safe and what it costs.
Still wondering?
help@personade.comLinkedIn has not said it scans private messages for AI writing. Since July 2026 members can flag posts with a "seems like AI slop" button, and LinkedIn says flagged posts get fewer views. For messages the judge is the person reading, and people spot templated wording in seconds. That is what this tool checks for.
A button LinkedIn added to posts on 30 July 2026 so members can flag content that reads as low-effort AI. It was used more than a million times in its first two weeks, according to LinkedIn's chief product officer as reported by Fortune.
It is a style check. It scores how much your text reads like AI or a template and shows you the phrases responsible. It does not claim to know who wrote it. Treat the score as a reader's first impression, not a verdict.
LinkedIn's help page says 200 characters, and free members can add a note to three connection requests a month. Some Premium and Sales Navigator accounts report 300. Writing to 200 means your note fits on every account.
Openers like "I hope this message finds you well" and "I came across your profile", phrases like "explore synergies", "unlock" and "elevate", long dashes, lists of three benefits, and praise that could fit anyone. One of these is fine. Four in a short note reads like a template.
No, not your message. We keep the score and a one-line summary with no names in it so your share link works, and the rewrite for 7 days in case you ask us to email it. Nothing you paste is used to train anything.
Yes. No account and no email needed to see your score, the flagged phrases and the rewrite. The email is only for the extra rewrite pack. There is a limit of ten checks an hour per person.
It removes the reason people ignore a message in the first second. Whether they reply after that depends on who you picked and whether you have a real reason to write. The tool fixes the wording; the targeting is up to you.
Further reading
- LinkedIn sequencesSend your connection note and follow-ups automatically, with each person's name in, and stop when they reply.
- How to write a cold message that gets readThe structure behind the rewrites above, with more examples.
- Outreach when it feels like spamHow to reach out to strangers without feeling like the thing you hate getting.
- LinkedIn account safetyDaily limits, warm-up and working hours, so sending more does not put your account at risk.
More free tools
All free tools- Coming soon
LinkedIn profile trust checker
Upload your own profile PDF or paste it in. See which signals make it look real or fake to LinkedIn and to the people you contact.
- Coming soon
LinkedIn profile audit
Will strangers accept and reply? Score your profile against the buyer you want, preview your invite card, and get a new headline and About opener.
- Free
Loom video downloader
Paste a public Loom link and keep the MP4. No account, no watermark.
Open the tool