ubiehealth.com — AI Site Grade

Analysis

Ubie's Corporate Site Is Invisible to AI Crawlers — While Its Consumer Content Is a Model for AI Readiness

The main consumer-facing site (Next.js, served via Fastly/Varnish behind istio-envoy) returns full HTML to every major AI crawler — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, all get 200 with ~117KB of content and zero UA-based blocking. But the corporate sub-site (/company/*, built on Studio.Design/Nuxt) returns zero visible text to the same bots — a pure JS shell with no server-side rendering.

Crawler Access

robots.txt allows all crawlers on / but disallows Next.js dynamic route patterns (/symptoms/[name], /diseases/non-predictable/[slug], /doctors/[slug], etc.) to save crawl budget. No AI-specific user-agent rules exist — no GPTBot, ClaudeBot, or Google-Extended directives at all. llms.txt returns 404. The sitemap contains 1,574 URLs — mostly symptom pages — and is referenced in robots.txt. The homepage sets cache-control: private, no-cache, no-store and a set-cookie header, which may cause repeated re-fetches by crawlers.

Content & Schema

The consumer site is rich in structured data. Symptom pages (e.g., /symptoms/headache) carry MedicalWebPage, FAQPage, and BreadcrumbList schema — including named physician reviewers with affiliations, dateModified, and full question-answer pairs. The homepage and most subpages have zero JSON-LD — no Organization, WebSite, or MedicalWebPage schema on the root. The Doctor's Note section has no schema at all. The corporate site (/company, /company/about, /company/policies) is entirely JS-rendered with zero extracted text — bots see only a <script> tag shell.

Cold-Knowledge Gap

The LLM knows Ubie as a "Japanese health tech company" founded in 2017 by two doctors, with a symptom checker and enterprise triage tools, partnerships with Japanese institutions, and funding from Norwest Venture Partners. The site itself presents Ubie as a US-facing consumer brand — "Developed by doctors," 50+ international supervising physicians, 1,700+ medical institutions, Newsweek "World's Best Digital Health Companies 2024," and a 71.6% top-10 accuracy score. The cold knowledge misses the US consumer positioning entirely — the accuracy benchmark, the Google Play "Best With AI" award, the Doctor's Note Q&A product, and the Smart Support enterprise triage product. The corporate founding story (two doctors, Tokyo 2017) is known but the site's About page is unreadable by bots.

External Signals

The site references partnerships with Google Health, IDEO, and Mayo Clinic Platform (on the Smart Support page). External links point to PubMed references on symptom pages and to recruit.ubiehealth.com for careers. No Reddit threads, press mentions, or review-site citations surfaced in search — the brand has minimal off-domain footprint in English-language search results.

Surprising Findings

The corporate subdomain is effectively invisible to AI — /company/about, /company, /company/policies all return <title> and meta tags but zero body text to any crawler. This means Ubie's own origin story, leadership, funding history, and policy pages are unreadable by GPTBot, ClaudeBot, PerplexityBot, and Google-Extended. Meanwhile, the consumer symptom pages are among the most AI-friendly medical content observed — rich schema, physician-attributed content, FAQ markup, and full HTML. The disconnect is stark: the pages AI engines would most want to cite for brand credibility are the ones they cannot read.

Findings

1. Corporate sub-site renders as empty JS shell to AI crawlers High

   Pages under /company/ (about, policies, company home) return zero visible text to GPTBot, ClaudeBot, PerplexityBot, and Google-Extended. Bots see only a script tag shell with no server-side rendering.

   What to change: Implement server-side rendering or static generation for all corporate pages so that AI crawlers receive full HTML content.

2. No llms.txt file published Medium

   The site returns a 404 for /llms.txt, missing an opportunity to guide AI crawlers to key pages and provide a structured summary of the site's content.

   What to change: Create an llms.txt file listing important pages (symptom checker, doctor's note, about, etc.) with brief descriptions.

3. No AI-specific user-agent rules in robots.txt Low

   The robots.txt does not include any directives for GPTBot, ClaudeBot, Google-Extended, or other AI crawlers. While all crawlers are allowed on /, the lack of explicit rules may lead to inefficient crawling or future blocking.

   What to change: Add explicit allow/disallow rules for AI crawlers to manage crawl budget and signal AI-friendly pages.

4. Homepage and most subpages lack JSON-LD schema Medium

   The homepage and many subpages (e.g., /how-ubies-ai-works, /clinical-data-sources) have zero JSON-LD structured data. No Organization, WebSite, or MedicalWebPage schema is present on the root, reducing AI understanding of the brand and site structure.

   What to change: Add Organization, WebSite, and MedicalWebPage JSON-LD schema to the homepage and key landing pages.

5. Doctor's Note section has no structured data Medium

   The /doctors-note page, which contains Q&A content, lacks any FAQPage or Article schema, missing an opportunity to be surfaced as rich results in AI and search.

   What to change: Add FAQPage or Article JSON-LD schema to the Doctor's Note pages.

6. Homepage cache-control headers may cause repeated crawls Low

   The homepage sets cache-control: private, no-cache, no-store and a set-cookie header, which may cause AI crawlers to re-fetch the page frequently rather than caching it, wasting crawl budget.

   What to change: Allow caching for AI crawlers by removing no-cache/no-store directives or using a separate cache policy for bots.

7. LLM cold knowledge misses US consumer positioning Medium

   The LLM knows Ubie as a Japanese health tech company but lacks awareness of its US-facing consumer brand, including the 71.6% accuracy score, Google Play award, Doctor's Note product, and Smart Support enterprise triage. This gap is partly due to the invisible corporate site.

   What to change: Ensure the corporate site is crawlable and contains key brand achievements, and consider publishing a press page or funding announcements in a crawlable format.

8. Minimal off-domain English-language footprint Medium

   Web searches for Ubie Health funding, reviews, and Reddit mentions returned zero results. The brand has little external citation or discussion in English, limiting AI knowledge beyond the site itself.

   What to change: Encourage press coverage, user reviews, and community discussions to build off-domain signals.

9. Smart Support page has very little visible text Low

   The /smart-support page returns only 10 words of visible text to a normal browser, and while bots receive full HTML (57KB), the page's content is minimal, limiting its value for AI understanding.

   What to change: Expand the visible content on the Smart Support page to describe the product more thoroughly.

What's working

• Symptom pages carry MedicalWebPage, FAQPage, and BreadcrumbList schema — Pages like /symptoms/headache include detailed JSON-LD with physician reviewers, dateModified, and full FAQ Q&A pairs, making them highly AI-friendly and eligible for rich results.
• Consumer site serves full HTML to all major AI crawlers — The main site returns ~117KB of rendered HTML to GPTBot, ClaudeBot, PerplexityBot, and Google-Extended with no UA-based blocking, ensuring all consumer content is accessible.
• robots.txt allows all crawlers on root and references sitemap — The robots.txt permits all user agents to access / and includes a sitemap reference, providing a baseline level of crawl access.
• Sitemap contains 1,574 URLs covering symptom pages — The sitemap lists a large number of symptom and disease pages, helping crawlers discover the site's content efficiently.
• Symptom content attributed to named physicians with affiliations — The symptom pages include physician reviewers with names and affiliations, adding credibility and transparency that AI systems can cite.
• Site references partnerships with Google Health, IDEO, and Mayo Clinic Platform — The Smart Support page mentions collaborations with reputable organizations, which can enhance AI trust and citation value.
• Clinical Data Sources page provides transparency — The /clinical-data-sources page explains the medical data behind the symptom checker, which can be used by AI to assess reliability.
• Doctors page lists physician contributors with content — The /doctors page provides 765 words of text about the doctors who develop Ubie, adding credibility and human context.