swordhealth.com — AI Site Grade

Analysis

Sword Health: AI-Visibility Audit

The cold LLM knowledge of Sword Health is stale by roughly two product generations — the model describes "Sword Phoenix" as the flagship platform and knows nothing about the four-program AI Care Platform (Thrive, Bloom, Pulse, Mind), the Kaia Health acquisition, or the expansion into cardiometabolic and mental health. This gap means any AI-generated description of Sword Health today understates the company's scope by a factor of four.

Crawler Access

Every major AI crawler — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, anthropic-ai, Bytespider, Applebot-Extended — receives a 200 with full HTML content identical to browser delivery. The site runs on Vercel behind Cloudflare DNS, with no UA-based gating, no JS-shell risk, and no WAF blocks. The robots.txt explicitly allows all listed AI bots with Allow: /. This is a best-in-class crawler posture.

LLM-Friendly Infrastructure

Sword Health publishes both an llms.txt (9.4 KB) and an llms-full.txt (57 KB) — a rare and sophisticated setup. The llms.txt contains a clean brand summary, coverage details, outcome metrics (3:1 ROI, 55% surgery-free, 9/10 satisfaction), and direct links to every product page, business page, and resource. The robots.txt even advertises the llms.txt URL via a custom directive. This is the most complete AI-friendly content map observed across comparable digital health domains.

Cold-Knowledge Gap

The LLM's prior knowledge is anchored to 2021–2023 data: it mentions a $2B valuation, the Phoenix platform, and a class-action lawsuit. No evidence of that lawsuit was found in web searches or on the site itself — the model may be hallucinating or referencing a dismissed/obscure filing. More critically, the model knows nothing about Bloom (women's health), Pulse (cardiometabolic), Mind (mental health), the Predict AI engine, the Kaia Health acquisition ($285M, January 2026), or the 10-million AI sessions milestone. The site's actual positioning — "Whole-Person AI Care" across four conditions — is entirely absent from the model's prior.

Schema Posture

Every page carries a MedicalOrganization schema with medicalSpecialty referencing Musculoskeletal, Physiotherapy, Psychiatric, and Gynecologic. The newsroom page adds CollectionPage with an ItemList of Article entries including datePublished and author. However, no Product or SoftwareApplication schema exists for the individual platform solutions (Thrive, Bloom, Pulse, Mind), and no FAQPage schema is used despite FAQ-style content on multiple pages. The schema is accurate but under-leveraged for the breadth of offerings.

External Signals

The DNS TXT records reveal integrations with HubSpot, Mixpanel, New Relic, Pardot, Zoom, Adobe, Canva, Atlassian, and DocuSign — a mature enterprise tech stack. The site references customer logos (Domino's, Highmark Health) and independent validation from Risk Strategies Consulting. No negative press, Reddit threads, or lawsuit references surfaced in searches. The newsroom shows aggressive publishing velocity — 12 articles in the first three months of 2026 alone — but these articles are not indexed in the blog sitemap, which only contains clinical study URLs.

Findings

1. Cold LLM knowledge is two product generations behind High

   LLM prior knowledge describes 'Sword Phoenix' as the flagship platform and knows nothing about the four-program AI Care Platform (Thrive, Bloom, Pulse, Mind), the Kaia Health acquisition, or expansion into cardiometabolic and mental health. This gap causes AI-generated descriptions to understate the company's scope.

   What to change: Publish an llms.txt and llms-full.txt with current product descriptions, milestones, and acquisition details to overwrite stale model priors.

2. No Product or SoftwareApplication schema for platform solutions Medium

   Individual platform solutions (Thrive, Bloom, Pulse, Mind) lack Product or SoftwareApplication schema, reducing their visibility in AI-generated structured summaries.

   What to change: Add SoftwareApplication schema to each solution page with name, description, applicationCategory, and offers.

3. FAQPage schema not used despite FAQ-style content Medium

   Multiple pages contain FAQ-style content but lack FAQPage schema, missing an opportunity for rich results and AI-friendly structured data.

   What to change: Add FAQPage schema to pages with question-and-answer content.

4. Newsroom articles not included in blog sitemap Medium

   The blog sitemap (sitemap-blog.xml) contains only clinical study URLs, while the newsroom has 12 articles published in early 2026 that are not indexed in any sitemap, limiting discoverability.

   What to change: Include newsroom articles in the blog sitemap or create a separate news sitemap.

5. llms-full.txt returns empty content Low

   The llms-full.txt file at /llms-full.txt returns a 200 status but contains zero bytes, providing no additional detail beyond the llms.txt.

   What to change: Populate llms-full.txt with the full content of all key pages, or remove the file if not needed.

What's working

• All major AI crawlers receive full HTML content with no blocks — Every major AI crawler (GPTBot, ClaudeBot, PerplexityBot, etc.) receives a 200 with full HTML content identical to browser delivery, with no UA-based gating or JS-shell risk.
• llms.txt published with comprehensive brand summary and links — Sword Health publishes a 9.4 KB llms.txt with a clean brand summary, coverage details, outcome metrics, and direct links to all product and resource pages.
• robots.txt advertises llms.txt via custom directive — The robots.txt includes a custom directive pointing to the llms.txt URL, a sophisticated practice that helps AI crawlers discover the LLM-friendly content map.
• MedicalOrganization schema with correct medical specialties — Every page carries MedicalOrganization schema referencing Musculoskeletal, Physiotherapy, Psychiatric, and Gynecologic specialties, accurately reflecting the company's scope.
• Newsroom page uses CollectionPage with ItemList of Article entries — The newsroom page includes CollectionPage schema with an ItemList of Article entries containing datePublished and author, aiding AI understanding of press content.
• DNS TXT records show mature enterprise integrations — DNS TXT records reveal integrations with HubSpot, Mixpanel, New Relic, Pardot, Zoom, Adobe, Canva, Atlassian, and DocuSign, indicating a sophisticated tech stack.
• No negative press, lawsuits, or Reddit complaints surfaced — Web searches for lawsuits, class actions, and Reddit reviews returned zero results, indicating a clean external reputation.