For sites built with Bolt, Cursor, v0, Lovable or any React app

Your vibe-coded app looks great to people. Does AI search see a blank page?

Most AI crawlers don’t run JavaScript. Blanc shows what your page actually sends them and writes the prompt to fix it in your builder.

Free. No sign-up, no card. Any country. Results in seconds. ·

A magnifying glass moves across a web page, finds a hidden detail and marks it as verified.

Covers the crawlers behind

  • ChatGPT Search
  • Claude
  • Perplexity
  • Google AI Overviews
  • Bing and Copilot

The SPA blind spot

Your users see an app. Without rendering, AI crawlers see an empty div.

Most AI crawlers fetch the raw HTML and don’t run client-side JavaScript. Unless your host renders pages for them, facts and schema that live in React state leave answer engines almost nothing to read.

What your users see

A rich React app

  • Interactive UI with dynamic cards and filters
  • Fast client-side navigation between pages
  • Prices, locations and details on screen

What GPTBot and PerplexityBot see without rendering

An empty shell

<body>
  <div id="root"></div>
  <script type="module" src="/src/main.tsx"></script>
</body>

Google runs JavaScript before reading, so this can go unnoticed in search rankings while AI assistants skip the page. Some hosts, including Lovable, render pages for verified AI crawlers, and Blanc checks for that. Read what blocks AI crawlers.

A page can rank on Google and still be blank to AI.

Nothing looks broken to human visitors, so these problems sit unnoticed for months. Three of the most common, out of the six in our guide to what blocks AI crawlers:

A robots.txt rule
One line written years ago, or added by a security plugin, can block OAI-SearchBot or PerplexityBot from the whole site. Read robots.txt for AI crawlers.
Content that needs JavaScript
Most AI crawlers read the HTML your server sends and stop there. If your text loads after scripts run, they see an almost empty page, unless your host renders it for them.
Missing facts in the schema
Without your hours, prices or address in structured data, engines have to guess, and they tend to quote someone who spelled it out. Read LocalBusiness schema for AI answers.

Four checks in one audit.

Every audit runs all four at once, and each finding shows the evidence behind it.

Crawler access

Which of 12 AI and search crawlers robots.txt lets in, at your exact page. Each crawler follows its own rules, so we check them one by one.

Structured data

Whether your JSON-LD is valid and describes a specific kind of business, product or article.

Fact extraction

Whether the facts people ask about, such as hours, prices, payment methods and parking, are stated where engines can read them.

Information gain

Whether answers come first and carry specifics, like prices, times and numbers, instead of generic claims. We cannot measure originality from outside, so we do not score it.

Isn’t this just SEO?

Mostly, yes. Good SEO covers a lot of what AI assistants need, so if yours is solid, the fixes here usually take an hour. The audit checks the four places where AI assistants read a page differently from Google.

TopicGoogleAI assistants

Who is allowed in

GoogleSEO checks confirm Googlebot can crawl the page.

AI assistantsChatGPT, Perplexity and Claude send their own crawlers. Firewall and security settings can block them while Google gets through, often without the owner knowing.

JavaScript

GoogleRuns JavaScript before reading, so script-loaded text still counts.

AI assistantsMost read the HTML as sent. A page that ranks well can look nearly empty to them.

Ranking vs being quoted

GoogleRanks the page so people click through, even if the answer is three paragraphs down.

AI assistantsLift a sentence or two. If the first sentence under a question is filler, they quote a page that answered directly.

Facts in the schema

GoogleSEO schema targets rich results: ratings, prices, events.

AI assistantsAnswer practical questions from facts rich results ignore: payment methods, parking, walk-ins, age rules.

The easiest way to settle it: audit your strongest page. If it comes back clean, your SEO already has you covered. Test a page or read the full comparison, AEO vs SEO.

How it works

  1. Run the audit

    Paste your deployed URL. Blanc inspects the raw HTTP response, entity schema and fact density, the way an AI crawler does, without running JavaScript.

  2. Stack detection

    Blanc detects whether you are on a React or Vite SPA, Next.js, WordPress or Shopify, and ranks the fixes for that stack, critical blockers first.

  3. One-click prompt fix

    Click “Copy AI prompt”, written for your stack. Paste it into Lovable, Cursor or Claude Code, deploy, and re-run the audit to see your new score. The prompt includes complete JSON-LD and llms.txt built from your page.

Running ChatGPT ads? Check the landing page before OpenAI’s crawler does.

OAI-AdsBot access, redirects, tracking parameters and login walls, in one free check. What we check

What a finding looks like

Plain language, the reason it matters, and the change to make.

1
robots.txt blocks OAI-SearchBot and PerplexityBot from this pageHigh priority

These crawlers fetch pages for AI search answers. While they are blocked, ChatGPT Search and Perplexity cannot read this page.

robots.txt
User-agent: OAI-SearchBot
User-agent: PerplexityBot
Allow: /

Built a client’s site with Lovable, Bolt or Cursor? Show them every page is readable.

With the Founding Pass, scan a client’s site, fix it with one prompt, scan again after you publish, and send a report with your name on it that shows what changed.

Any single-page audit can become a client report too. Free reports carry a small Blanc credit line, and the Founding Pass removes it. Hide the code for owners who just want the summary, or leave it in for their developer.

See the Founding Pass

Managing many client sites? Join the Agency waitlist.

Scan the site
Every page we find, up to 25 per scan, one from each section first. On a bigger site, the result names the pages it didn’t check.
Fix it with one prompt
One prompt covers every issue on every page checked, written for the builder the site uses, with code built only from facts on each page.
Scan again after you publish
Each scan is a numbered run. Compare any two: what got fixed, what’s new and which scores moved.
Send the progress report
Your logo and colours, no Blanc credit line, and a “Progress since run 1” section your client can read. No promises of rankings for you to walk back later.

Where these rules come from

Built on our own directory of 100+ kid-friendly places in Metro Manila.

Before this tool existed, we ran into the problem ourselves on NearMe Kids PH, our parent-run directory. Parents ask AI assistants about opening hours, entrance fees and venue rules, and we needed those answers to come out right. Every check in this audit started as a fix we made there.

NearMe Kids PH is a client-side React app built with Vite. When we first audited its homepage, the raw HTML our scanner received held 9 words. Everything a parent needs appeared only after JavaScript ran, and it scored 57/100.

We kept the app as it was and added a fallback to index.html: a factual summary, three answered questions, 29 crawlable links and our full Organization schema, all readable before any script runs. The same audit now scores 99/100.

Our first version of that fix also added a homepage canonical tag that every page would have inherited. We caught it, removed it, and changed how this audit handles canonicals.

Later we found a blind spot of our own. The site's host, Lovable, serves verified AI crawlers a rendered version of each page, so those 9 words were our scanner's view, not ChatGPT's or Claude's. Blanc now checks that rendered version too, and the case study says so.

57 → 99audit score

9 → 339words in the raw HTML

0 → 29crawlable links

Read the full case study Then: how we pre-rendered all 144 pages

Facts the official sites leave out

Venue websites often put prices in image banners, leave hours out of their schema, or load details with JavaScript. Our listings spell those facts out in text and structured data, which is exactly what the audit now checks for.

Tested across 100+ real listings

Indoor playgrounds, museums, family cafés and parks each need different facts. Working through them is how we arrived at the operational checklist: hours, fees, payment methods, parking, age rules and accessibility.

Answers first, adjectives last

“A magical family paradise” tells a parent nothing. “₱350 for one hour on weekdays, grip socks required” answers the question. The audit scores pages on how quickly they get to facts like that.

ChatGPT lists our Quezon City guide among its sources (same prompt as the others; the question is above this view)
ChatGPT lists our Quezon City guide among its sources (same prompt as the others; the question is above this view)ChatGPT · 2026-11-04 · Query: “free kid-friendly places in Metro Manila”
Perplexity cites two NearMe Kids PH guides in its answer and sources
Perplexity cites two NearMe Kids PH guides in its answer and sourcesPerplexity · 2026-11-04 · Query: “free kid-friendly places in Metro Manila”
Gemini lists our free-places guide first among its sources
Gemini lists our free-places guide first among its sourcesGemini · 2026-11-04 · Query: “free kid-friendly places in Metro Manila”
Copilot cites nearmekidsph for free National Museum admission
Copilot cites nearmekidsph for free National Museum admissionMicrosoft Copilot · 2026-11-04 · Query: “free kid-friendly places in Metro Manila”
652 citations across Microsoft Copilot and partners in 3 months
652 citations across Microsoft Copilot and partners in 3 monthsBing Webmaster Tools, AI Performance · Aug to Nov 3, 2026

Hard for AI to use

A typical venue website

  • robots.txt blocks AI search crawlers, or a firewall turns them away
  • Generic WebPage schema with no hours, prices or amenities
  • Entrance fees inside a promotional banner image
  • Opens with “Welcome to our wonderful world of play…”

Easy for AI to quote

A page that passes the audit

  • OAI-SearchBot, PerplexityBot and Claude-SearchBot allowed
  • LocalBusiness schema with hours, price range, payment methods and amenities
  • Prices, hours and parking rules written right under the question they answer
  • Opens with “Entrance is ₱350 on weekdays. Grip socks are required.”

Passing the audit makes a page easier to read and quote. It does not guarantee a citation; what other sites say about you matters too.

When a parent asks an AI assistant a practical question, the page that states the answer plainly is the one that can be quoted. This tool checks any page for that in a few seconds.

Jeron Clemente, co-founder of NearMe Kids PH and creator of Blanc

Start free. Upgrade when you need more.

Free

$0

Up to 3 sites, 1 page each. Score, ranked fixes, the AI fix prompt and ready-to-paste code built from your page.

Founding Pass

$19 one-time

5 sites and 50 page audits for 12 months: site scans of up to 25 pages with an internal links check, scan history, and client reports with only your brand. 30 passes at the founding price.

Agency

Early access

Planned: scheduled checks with alerts, full crawls beyond 25 pages with a full link map, and bulk audits across client sites. Join the waitlist for early access.

Compare plans in detail

What we won’t promise

No tool can guarantee that ChatGPT or Google will cite you. These fixes remove the reasons an AI engine might skip your page, which is the part you control. The rest depends on what other sites, reviews and roundups say about you, and we will tell you that up front instead of selling you autopilot traffic.

Questions

Is the audit free?

Yes, for up to 3 sites, 1 page per site, with no sign-up and no card. Re-scanning a page you’ve already audited is always free. The Founding Pass covers 5 sites and 50 page audits, with site scans of up to 25 pages at once.

Does it work if my site isn’t vibe-coded?

Yes. Blanc audits any public page: WordPress, Shopify, Webflow, Wix, Squarespace, GoHighLevel or hand-coded HTML. The fix prompt is written for the platform it detects, with click-by-click steps where you can’t edit code.

How does the score work?

It adds up 4 parts: AI crawler access (25 points), schema and entities (30), answer-ready content (30) and technical readability (15). The weighting is Blanc’s own way to rank fixes, not an industry standard, and no AI company publishes a score like it. Every finding shows its evidence, so you can check it with free tools: View Source for the text in your HTML, your robots.txt file, and a schema validator for structured data. Google says its AI features need no special schema, so treat structured data as making your facts clearer, not as a Google requirement. If AI can read your site and the score is still low, read why a readable site can still score low, and what to fix first.

Isn’t AEO or GEO just SEO?

Mostly yes, with 4 differences. AI assistants use their own crawlers that firewalls can block while Google gets through, most of them do not run JavaScript, they quote a sentence instead of ranking a page, and they use practical facts like payment methods and parking that SEO schema rarely includes. The audit checks those differences.

Can you check if my landing page is ready for ChatGPT ads?

Yes. Run the ChatGPT Ads readiness check from the ChatGPT Ads page or the homepage. It checks robots.txt rules for OAI-AdsBot and OAI-SearchBot, the HTTP status and redirect chain, HTTPS, UTM tracking, login walls, CAPTCHAs and firewall challenges. It cannot guarantee approval, because our scanner is not OpenAI’s crawler and OpenAI also reviews the ad against its policies.

Will these fixes get my site cited by ChatGPT?

No, and no tool can promise it. The fixes make your page easier for AI engines to read and quote, but AI answers also depend on what other sites, reviews and roundups say about you.

Which AI crawlers does it check?

12 crawlers: OAI-SearchBot and ChatGPT-User from OpenAI, Claude-SearchBot and Claude-User from Anthropic, PerplexityBot and Perplexity-User, Googlebot and Bingbot. It also reports GPTBot and ClaudeBot, used for model training; Google-Extended, which covers Gemini training and Gemini Apps answers; and OAI-AdsBot, used for ChatGPT ad review. None of those four affect the score.

Does the audit run JavaScript?

No, on purpose. Most AI crawlers read the HTML your server sends without running scripts, so the audit sees what they see.

Can I run Blanc from an AI assistant?

Yes, if the assistant can open web addresses. Ask it to open https://heyblanc.com/audit?url=yoursite.com, with your page’s address after url=, and read what comes back. It is the same free audit as this page, as data the assistant can read: the score, the fixes in order, the code to paste in and the fix prompt. It needs no sign-up and no key. It allows 10 audits a minute, and a repeat request for the same page within 10 minutes returns the same result. Site scans and white-label reports need a Founding Pass on this site. Facts about Blanc for AI tools, including how to describe it, are on our page for AI assistants.

What code does the audit write?

3 things, built from your page: complete JSON-LD with the facts already on it, an FAQ block from questions you answer, and an llms.txt that lists your real pages. Anything it could not find is listed separately, so no example values end up on your live site.

Do you store the pages I audit?

10 minutes: results are cached that long so repeat checks are fast, then discarded. You do not need an account to run an audit, and the site shows no ads and uses no advertising cookies.

See what your page looks like to AI crawlers.

Run free audit