Why would an AI company score poorly on AI-search readiness?

Most AI startups ship a client-rendered single-page marketing site with little server-side HTML, sparse structured data, and sometimes no sitemap. AI search crawlers that do not execute JavaScript then see little content to classify or quote — independent of how good the company's own AI product is.

What was the most common gap?

Missing JSON-LD structured data and thin server-rendered content. Among 25 leading AI startups the average score was 83/100, but the laggards shared those two gaps; the lowest scorers also lacked a discoverable sitemap.

How do I check my own site?

Use the free checker at samedaydesk.com/tools/ai-readiness, or run npx github:epistemedeus/ai-readiness yourdomain.com

SameDayDesk · Report · June 2026

We tested 25 top AI startups for AI-search readiness. Perplexity scored a C. LlamaIndex scored a D.

AI engines like ChatGPT, Perplexity, and Google AI can only cite pages they can actually read. So we pointed an AI-search readiness check at the homepages of 25 well-known AI/LLM startups — the companies building this very future. Many do great. But some of the most recognizable names are surprisingly hard for AI search to read, including an AI search engine and a company whose product is making data legible to LLMs.

25AI startups scanned

83average score /100

7scored C or D

1scored a D (LlamaIndex)

The full results

Scored June 23, 2026 with the same free AI Readiness Checker anyone can run. Each homepage is checked for AI-crawler access (GPTBot, ClaudeBot, PerplexityBot, Google-Extended), JSON-LD structured data, title/meta, Open Graph, an XML sitemap, and llms.txt.

Company	Score	Grade
llamaindex.ai	53	D
character.ai	60	C
perplexity.ai	60	C
replicate.com	63	C
cerebras.ai	68	C
poolside.ai	68	C
browserbase.com	71	C
groq.com	76	B
harvey.ai	76	B
langchain.com	76	B
mistral.ai	78	B
weaviate.io	78	B
together.ai	83	B
modal.com	85	B
runwayml.com	90	A
contextual.ai	93	A
codeium.com	98	A
fireworks.ai	98	A
glean.com	98	A
cohere.com	100	A
cursor.com	100	A
elevenlabs.io	100	A
lovable.dev	100	A
pinecone.io	100	A
suno.com	100	A

The irony, and the lesson

Perplexity, an AI answer engine, scored a 60 (C). LlamaIndex, whose entire product is about making data readable by LLMs, scored a 53 (D) — the lowest in the set. This isn't a knock on their products; it's a reflection of how their marketing sites are built. The pattern among the C's and D's was consistent:

JavaScript-only rendering. A client-rendered SPA serves crawlers that don't run JS a near-empty shell. The product is great; the homepage an AI crawler sees is thin.
Little or no JSON-LD structured data, so engines have no machine-readable description to quote.
No discoverable sitemap on the lowest scorers.

The A-graders (Cohere, Cursor, ElevenLabs, Pinecone, Suno, Lovable, Glean) simply ship server-readable content, structured data, a clean robots.txt, and a sitemap. None of it is exotic — it's hygiene. And if the companies building AI are leaving citations on the table, almost every other business is too.

Check your site in 10 seconds

Run the free AI Readiness Checker on any URL. If it flags gaps, the Fix Pack ($39) ships a corrected robots.txt, sitemap, JSON-LD structured data, and clean title/meta/OG tags — installed-ready, today.

Run the free check Get the Fix Pack · $39

Method: each company's homepage was fetched once and scored on AI-crawler access, JSON-LD, title/meta, Open Graph, XML sitemap, and llms.txt. This is a point-in-time, homepage-level snapshot, not a full-site audit, and scores change as sites update. Run your own at samedaydesk.com/tools/ai-readiness.