Fact-check any LinkedIn post in one click.
Every post in your feed gets a small button. Click it, pick an action, and the post opens in a new tab at the AI service you use, with your prompt already typed. The extension has no server and no account. It reads nothing until you click, and nothing it reads ever reaches me.
Companies that adopted AI in 2025 saw an average 340% ROI within six months. If your team is still "evaluating", you've already lost. Agree?
Five manual steps, collapsed to one.
Copy the text, open a chat, paste, write the prompt, read the answer. Reposting takes one tap, so unchecked claims win by default. The extension evens the odds.
- 1
You click a post's button
The extension runs on linkedin.com only. Until you click, it reads nothing and sends nothing, and it never looks at your profile, messages or connections.
- 2
It reads that one post
The author's name and headline, the post text, links to its images, and whatever comments are on screen. Often the correction you need is already in the replies.
- 3
A tab opens at your AI
The post and your prompt go into the URL of a new tab at the service you picked. From there it is the same as if you had pasted it yourself. Same account, same terms.
Whichever one you already pay for.
- ChatGPT
- Claude
- Perplexity
- Grok
- Microsoft Copilot
Pick one in the options page. The answer lands in the account you already have, at the service's own address, and the extension has no part in it after the tab opens.
The whole policy, in three lines.
- Collects
- nothing. No server, no analytics, no crash reports, no account.
- Stores
- your settings, in your browser, on your device. Uninstall and they are gone.
- Sends
- the post you clicked, to the AI you chose, when you click. Never to me.
The prompts are yours to write.
Every menu action is a name and a prompt, editable in the options. Keep the ones it ships with, rewrite them, or add your own. A few worth having:
Fact-check this post
Check every factual claim and number in this post. For each one, say whether it is supported, cite the primary source if one exists, and flag anything that is misattributed, outdated or invented.
Find the primary source
The post quotes a statistic. Find the original study or report it comes from, tell me what the study actually measured, and note where the post's version differs.
Summarize the comments
Summarize the comments below the post. Surface any that correct or contradict the post, and any that add a source.
Explain it plainly
Rewrite this post in plain language, removing the hype. Then tell me in two sentences what, if anything, is worth taking from it.
My feed was full of numbers nobody had checked.
I read LinkedIn for work. Most of what ranks there is a confident statistic with no source, and the same figure keeps circulating years after the study behind it was misread or retracted. I was checking claims by hand a few times a week, and the friction meant I mostly didn't bother. So I wrote the extension that removes the friction and left my own judgment in the loop.
It is also a small demonstration of how I work. One person, an evening, a product that ships with no backend to maintain and nothing to leak. The same AI-native way of building I install in client teams, at the scale of a browser extension. It sits beside Watchly on the shelf of products I own and run.
What people ask about the extension.
What is LinkedIn AI Extension?
LinkedIn AI Extension is a Chrome browser extension built by Oshri Cohen that adds a button to every post on linkedin.com. Clicking the button opens a menu of actions, such as fact-checking the post, and each action sends the post's text, author line, image links and visible comments, together with a prompt, to the AI service the user selected: ChatGPT, Claude, Perplexity, Grok, Microsoft Copilot or Google.
Does the extension collect any data?
No. The extension has no server, no analytics, no crash reporting, no advertising and no account system. The developer receives no data from it. Settings are stored in the browser's local extension storage on the user's device and are deleted when the extension is uninstalled.
Where does the post text go when I click an action?
The text read from the post is combined with the prompt and placed in the URL of a new browser tab opened at the AI service selected in the extension's options. From that point the data is handled by that service under its own privacy policy, exactly as if the user had pasted it into the chat. The extension never sends data to the developer or to any other third party.
What permissions does it request?
Two. Access to pages under linkedin.com, to add the button to posts and read a post when clicked, and the storage permission, to save settings on the device. It does not request access to other websites, browsing history, tabs, cookies or the clipboard, and it does not download or run remote code.
Can I write my own prompts?
Yes. The menu actions and their prompts are editable in the extension's options page. Users can keep the defaults, edit them, or add actions such as finding the primary source of a claim, summarizing the comment thread or translating the post.
Why did Oshri Cohen build it?
Oshri Cohen built LinkedIn AI Extension because unsourced statistics spread on LinkedIn faster than anyone checks them, and checking took five manual steps against one tap to repost. The extension collapses those steps into one click while keeping the user's own AI account and judgment in the loop. It is also a small, working example of his AI-native thesis: one person shipping and running a product end-to-end.
Check the claim
before it becomes yours.
Questions about the extension, the privacy policy, or how one person ships and runs tools like this? Email me. I answer them myself.