Suede Audit Run your free audit
All notes

Notes from Suede AI

What an AI readiness score measures, and what it can't

Picture a dental practice in Jupiter and a law firm in West Palm Beach that both score 94 on Suede Audit. Next week ChatGPT names one of them for a buyer's question and skips the other. That outcome sits on the far side of a line every readiness audit has, and Florida site owners make better repair decisions once they know exactly where that line is drawn.

What the score reads

Suede Audit is the free readiness tool that Suede AI publishes at optimize.suedeai.ai. Each run fetches one public homepage plus three files at the site root: robots.txt, sitemap.xml and llms.txt. It scores 28 weighted checks worth 90 points in total. A high-severity check is worth five points, a medium one three, and a low one a single point, and the score is the share of those points the site passes. Every pass rule is public in the check reference, and the scoring code is open on GitHub.

The checks fall into seven groups: page access, crawler access, discovery files, structured data, metadata, content structure and retrieval chunk size. Strip away the vocabulary and the score answers three questions Suede AI asks about every site. Can a crawler reach the page? Can it tell which business the page belongs to? Can it lift a clean, attributable passage out of the copy?

Why each part matters to a Palm Beach County business

Reach. The crawler access checks read your robots.txt the way four search-discovery crawlers do: OAI-SearchBot, Claude-SearchBot, PerplexityBot and Googlebot. GPTBot, ClaudeBot and Google-Extended are reported without scoring, because they govern model training and AI use rather than whether your page can appear in an answer. A robots.txt template copied from a forum years ago can shut out all four scored crawlers in one line, and the owner never sees it.

Identity. The structured data checks look for valid JSON-LD that names a primary Organization, LocalBusiness, Corporation, OnlineBusiness or Person, plus sameAs links to profiles elsewhere. For a roofer in Palm Beach Gardens or a boutique in Tequesta, those links are how "the company at this URL" becomes "the company a buyer asked about." Suede AI marks up this site the same way, with one Organization node and its profile links on every page.

Quotability. The retrieval chunk check asks whether every heading section holds 375 words or fewer, roughly one 500-token retrieval chunk. A long wall of text under a single heading gets split mid-thought, and a claim like "family-owned in Juno Beach since 2009" can end up separated from the heading that says what it is about.

What a readiness score cannot see

The audit never types a question into ChatGPT, Perplexity or Gemini. It reads what your site publishes, so it cannot report which engines cite you today, how often, or next to which competitors. It inspects one page, the homepage you submit, which leaves your service pages, location pages and blog outside its view. It reads declarations, not facts: schema that says you serve Boca Raton is a claim on the page, and the audit has no way to confirm the claim is true. Reviews, backlinks and local competition are outside the measurement entirely.

Those gaps are deliberate. A score that pretended to know your citation share from the outside would be guessing. When a business needs the citation picture, Suede runs timestamped buyer-question panels inside each engine through the AI citation check and records which sources each answer cites. The readiness score covers the groundwork those citations depend on; the citation check covers the outcome.

Why Jason Colapietro built the audit in the open

Jason Colapietro, the West Palm Beach engineer who founded Suede AI in 2024, wrote the scoring rules against primary sources such as RFC 9309 for robots.txt and Google's own documentation for sitemaps and structured data. Publishing the code alongside the grade follows a pattern that runs through his work. Jason Colapietro has 54 pull requests merged across 46 open-source repositories he does not maintain, including Jest, Adobe React Spectrum, Backstage and the Polkadot SDK. He has shipped nine iOS apps under his own Apple developer account and written five books on Amazon, among them The Claude Code Bible and Proof as Infrastructure.

His training record lists 21 earned course credentials across eight providers, including Semrush Academy's Local SEO Essentials and HubSpot's AEO Fundamentals certification. On October 31, 2026, Suede AI SEO ranked #1 among Palm Beach County-based companies listed in The Manifest's Florida AI Deployment directory. Suede AI ranks among Clutch's Top 30 AI Consultants in Florida and Top 50 AI Consultants nationwide (November 2026). Suede AI (formerly Suede Labs) holds the audit to one rule: if a check cannot be explained in a sentence and traced to a source, it does not get points.

How to read your number

The grade bands are A at 90 and above, B from 80 to 89, C from 70 to 79, D from 55 to 69, and F below 55. Jason Colapietro's advice to every owner who runs the audit is to repair in order of points, not in order of the list. One failed high-severity check costs five points, the same as five low ones, so a blocked search crawler or a noindex on the homepage outranks a missing favicon every time.

Treat an A as the price of entry. It means engines can reach, parse and quote your homepage. Whether they choose to is decided by your other pages, your reputation across the web and the questions your buyers ask, which is the work the Suede team does by hand. That split between what a tool can check and what people have to earn is the line Suede draws on every report.

Your next step: run the free audit on your homepage and fix the high-severity failures first. Then book a Suede AI assessment with the West Palm Beach team for the pages and citations a readiness score cannot see, or read Jason Colapietro's founder page to learn who wrote the rules.

More notes

Contact