2026-08-03 · Jack Stovell
Can AI read your Notion page, Substack, Medium — or your PDF CV?
Whether AI can read your page comes down to two questions: does the server send real HTML without needing JavaScript, and does the platform's robots.txt let AI crawlers in. By that test, GitHub profiles, Substack posts and published Framer sites read cleanly; public Notion pages and most of Medium do not; and a PDF CV is readable when you hand it to an assistant, but weak as a hosted identity source. Here is each platform, tested rather than assumed.
Can AI read a public Notion page?
Mostly no. Fetch a public notion.site page the way AI crawlers fetch — raw HTML, no JavaScript execution — and the only visible text in the response is a notice that JavaScript must be enabled, under a generic page title. We ran exactly that fetch while writing this guide (August 2026): none of the page's actual content is in the HTML. Most AI crawlers do not execute JavaScript, so to them your carefully built Notion homepage is a JavaScript shell. Note that Notion's search-indexing toggle governs whether engines are invited to index the page — it does not change how the page renders. If a Notion page is your main public presence, this is a real gap.
Can AI read your Substack?
Largely yes. Substack posts are server-rendered HTML on the open web: free posts and archives arrive as readable text with no JavaScript required. Two caveats. Paid-subscriber posts sit behind the paywall and are as unreadable to machines as to logged-out humans. And robots rules vary between publications — when we checked in August 2026, one publication's robots.txt blocked GPTBot entirely while another large one did not, and Substack exposes signals like ai-train=no on some domains. So the platform is friendly, but check yourdomain/robots.txt rather than assuming.
Can AI read your Medium posts and profile?
Less than you would expect. Medium's pages are server-rendered, but medium.com's robots.txt — as of our check in August 2026 — disallows nearly the entire site to a named list of AI crawlers including GPTBot, ClaudeBot, Amazonbot, Applebot-Extended, Bytespider and meta-externalagent. Ordinary search crawlers are still allowed, so fragments of your writing can reach AI answers indirectly via search indexes, but a direct fetch by a robots-respecting AI crawler is turned away, and member-gated posts are paywalled on top. Publishing on Medium is fine; making Medium your canonical home is not.
Can AI read your GitHub profile README?
Yes, cleanly. A GitHub profile page is plain server-rendered HTML with your README content right in the page, and profile URLs are not blocked in GitHub's robots.txt. For developers this is one of the strongest free identity pages available: a stable URL, real content, and links out — use it to point at your site or profile so machines can connect your accounts.
Can AI read your PDF CV?
It depends where the PDF is. Uploaded into a chat, yes: assistants parse an attached PDF well, and that is the designed use. Hosted at yourdomain.com/cv.pdf, it is weak as an identity source. Crawlers and retrieval systems overwhelmingly prefer HTML pages; a PDF carries no schema markup, no canonical URL semantics, and a designed two-column CV can extract in scrambled reading order. A plain web page stating the same facts will beat the PDF as the thing retrieval finds — keep the PDF for humans who ask, and give machines a page built to be read.
Can AI read a Framer site?
Yes. Framer pre-renders published pages to HTML at publish time and serves that HTML from a CDN — Framer's own hosting documentation describes this — so the content is present before any JavaScript runs. The usual caveats still apply: content that only appears through client-side embeds or interactions stays invisible, and your robots.txt stance matters here as it does everywhere.
What's the pattern?
Server-rendered public HTML wins; logins and JavaScript shells lose. That single rule predicts every verdict above, and the landscape moves — platforms change renderers and robots files without announcements, so today's verdicts have a shelf life. So test the page you actually care about rather than trusting a platform's reputation: here is how to check what AI can read on any site, and our free checker at hilyt.it/tools/ai-visibility runs the fetch, rendering and robots tests in one pass.
Check your AI visibility — free
Related guides
- JS shells and soft 404s: how your site lies to crawlers
- How to check whether AI can actually read your website
- How to create an AI-readable profile of yourself