2026-08-01 · Jack Stovell
Why LinkedIn can't get you cited by AI
Here is an uncomfortable experiment. Open a private browser window — no cookies, no session — and visit your own LinkedIn profile. What you get is not your profile. It's a sign-in wall, or a teaser card with most of your history missing.
That logged-out view is roughly what an AI system gets when someone asks it about you and it goes looking.
The wall is the whole problem
When you ask ChatGPT, Claude or Perplexity about a person, the tool usually runs a live web search and reads the pages that come back. It reads them the way a crawler does: no account, no cookies, often no JavaScript execution.
LinkedIn's business depends on keeping profile data inside LinkedIn. That's a legitimate choice — but it means the canonical professional profile of the last decade is structurally unreadable to the systems people increasingly ask instead of Google.
The result: your most complete, most current professional record contributes almost nothing to what an AI says about you.
"But I show up in Google"
Google has commercial arrangements and a decade of special-cased indexing around big platforms. AI retrieval tools mostly don't. Ranking in Google for your name and being legible to an AI answer engine are different problems, and the second one is the one growing.
There's a second trap here: even when a search tool finds your LinkedIn URL, a link is not knowledge. If the page behind it won't open, the model is left to guess from the URL and whatever scraps rank alongside it. Guessing is where wrong answers come from.
What the AI actually reads instead
Whatever is publicly crawlable with your name on it. For most working professionals that's a thin, stale mix: a conference bio from four years ago, an old employer's team page, a namesake's profile. The model doesn't know which of it is you, current, or true — so it hedges, or worse, blends.
What to do about it
You don't need to leave LinkedIn. You need one page that isn't LinkedIn:
- Publicly readable: no login, no JavaScript required to see the content.
- Explicit about identity: schema.org Person markup, with sameAs links to your site, GitHub and socials so models can tell you apart from namesakes.
- Dated and sourced: claims with years and links get repeated by models; vague prose gets hedged.
- Fresh: a visible last-updated date settles conflicts between old and new information.
Then link that page from the profiles you already have — including LinkedIn's website field — so retrieval finds it next to your name.
This is precisely the page a hilyt profile is: open HTML for crawlers, identity markup, sourced claims, and a plain-text llm.txt card any model can read in one pass.