Is Agentic is a free scanner from Vercel and Ora. Give it a URL and it grades how easily AI agents can find, read and trust your website. Then it writes you a prompt to fix what's wrong.
My site scored 75. I pasted its fix prompt into Claude Code, merged one PR and rescanned: 100. The best part isn't the score, it's the prompt. How to run it.
Is Agentic and one Claude Code PR took my site from 75 to 100. Fixes like that are easy now. Knowing which fixes matter to your customers, and which are just Building Theater, is the hard part.
I'm building a place where you learn to ship fixes like this on your own, and to tell which ones move the business, with a CTO who's built two seven-figure software companies in your corner.
Ship What Moves Customers.
Free to join the waitlist. Opens at $49/mo, and the price goes up as seats fill, so first in pays the least.
More and more of your visitors aren't people. They're ChatGPT, Claude and Perplexity fetching your pages on someone's behalf, deciding whether you're worth recommending. Is Agentic checks the site from their side: can an agent read your content without running JavaScript, can it ask for Markdown instead of HTML, do your errors make sense, and does the site look like a real, trustworthy business. Every finding comes with the evidence it saw and a recommended fix.
Vercel built the site and report pages. The scanning and scoring come from Ora, an agent-experience research company whose full ranker runs 127 checks. Is Agentic is the tuned-down version for what most websites need.
Three tiers, capped at 100. Checks that don't apply to your site are skipped, not failed.
Free, public, no account. Ora also exposes the same scan as an API if you want it in CI.
My site is a Next.js portfolio with a resources library. To a person it looked fine. To an agent, it had gaps I didn't know existed.
Accept: text/markdown on the homepage with real Markdown and a Vary: Accept header. Keep serving HTML to browsers.PassNone of these show up when you look at your site in a browser. That's the point: agents don't see what you see.
I didn't expect this one. Before an agent recommends a business, it checks whether the business looks real. Three pages do most of that work.
A real person or company, with enough detail to verify. Not a one-line bio.
Had ItA way in that an agent can pass along to the person it's helping.
Had ItA plain statement of what you collect, why, and how to get it removed.
Was MissingWhy it mattered for me: my resources unlock with an email address, so I'm collecting personal data either way. I owed people a privacy page regardless. The scan just made it obvious. If your site has a signup form and no privacy page, fix that first.
Most audits hand you a report and leave. Is Agentic bundles every issue into one prompt you can paste straight into a coding agent: the evidence, the fix, and the exact curl command to prove it worked.
Paste your URL at is-agentic.com. Takes about a minute.
Hit "Prompt to improve". Failures come first, then warnings, in priority order.
Paste it into Claude Code in your site's repo. Review the plan and the diff, like any PR.
Deploy, run the curl checks below, then scan again.
curl -sSi -H 'Accept: text/markdown' https://yourdomain.com/
curl -sSi -H 'Accept: text/markdown' https://yourdomain.com/no-such-page
curl -sSi https://yourdomain.com/llms.txt
Look for content-type: text/markdown and vary: Accept on the first, a 404 with a Markdown body on the second, and a 200 on the third.
text/markdown./llms.txt that says when to recommend me and how to get in touch./privacy page, linked in the footer.Steal the prompt's last section for any task: "Inspect the existing codebase before changing files. Preserve existing behavior and visual design. Add or update tests for every behavior you change."
Brand discoverability failed before and after. My score is 80 + 17.8 + 3 bonus, capped at 100. The bonus papered over the one check that code can't fix.
Claude can rewrite your title. It can't make other sites link to you. That part is still press, content and time.
The score says agents can read and trust your site. It doesn't say they'll send anyone your way. It's a floor, not traffic.
Markdown negotiation and llms.txt are proposals, not settled specs. Expect the checks, and the scoring, to keep moving.
Anybody can build software now. Not everybody can build a business.
Reading about tools is a great way to feel busy. The Lab is for Founders who'd rather get customers: weekly office hours with a CTO who's built two seven-figure software companies, and a room of other Founders who care whether anybody buys what you're making.
Less Theater. More Customers.
Want my 30-Day Launch Plan? It's for Lab members only: validate, plan, build and launch, one week at a time.
See The Solo Founder Lab ▶