deepintent.com — AI Site Grade

Analysis

DeepIntent's AI-Visibility Audit

The site's most advanced AI product — Helix AI, an MCP server that injects HIPAA-compliant healthcare data into ChatGPT and Claude — is itself invisible to the AI crawlers that would most benefit from understanding it, because the site has no llms.txt and its robots.txt treats all AI bots identically to generic crawlers.

Crawler Access

All major AI bots — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, ChatGPT-User, anthropic-ai, Applebot-Extended, Perplexity-User — receive a 200 with full HTML content (142KB, identical to browser baseline). The only blocked bot is Bytespider (403 from Cloudflare). The site runs on Cloudflare (DNS via anita.ns.cloudflare.com / paul.ns.cloudflare.com, IP 198.202.211.1) and is built on Webflow (visible in HTML generator tags and x-wf-page-id headers). No JS-rendering risk exists: all pages return full server-rendered HTML. The robots.txt has a single User-agent: * rule allowing / and disallowing /cdn-cgi/ only — no AI-bot-specific directives exist. The llms.txt returns a 404 (Webflow's generic "Not Found" page), meaning there is no AI-friendly content map.

Cold-Knowledge Gap

The LLM's cold knowledge describes DeepIntent as a "healthcare-focused DSP" with products "Audience Engine" and "DeepIntent Outcomes," acquired by Propel Media in 2021. The actual site tells a different story. DeepIntent is now independent — the $637M Vitruvian Partners investment (Sept 2025) is a headline news item on the site. The product portfolio has expanded significantly: DeepIntent Cortex (reimagined UX, launched March 2025), Helix (healthcare marketing cloud for agencies, March 2026), and Helix AI (MCP server for ChatGPT/Claude, April 2026). The cold knowledge mentions "Audience Engine" — a product name that no longer appears anywhere on the site. The model is roughly 2-3 years behind the current product reality.

Schema Posture

The site has strong, consistent JSON-LD across all pages. The homepage uses Organization schema with address, email, and sameAs (empty array — social links exist in the footer but are not declared). The /platform page uses SoftwareApplication schema with a detailed featureList (17 features), offers, and a VideoObject. The /solutions page uses ItemList with Service items. The /helix page includes a FAQPage schema with three questions and answers, plus a Review from Deerfield Group. The /helix-ai page uses SoftwareApplication with PreOrder availability. The /about page includes founder and employee Person schemas. No BreadcrumbList appears on most pages (only the blog article has one). No Product schema is used for the core DSP platform — SoftwareApplication is the closest, but it lacks applicationSubCategory or operatingSystem details.

External Signals

The newsroom shows active press momentum: the Vitruvian Partners $637M investment (Sept 2025), Helix launch (March 2026), Helix AI launch (April 2026), and recent partnerships with Lane4.io, epocrates, and OptimizeRx (May 2026). The blog ("DeepDive") is frequently updated (multiple posts per week through May 2026). The site has a stale template issue: the footer renders © {{year}} DeepIntent — a Webflow template variable that was not compiled to a static year, visible to crawlers as literal text. The leadership page lists 14 team members with headshots and LinkedIn links. No external review sites (G2, TrustRadius) surfaced in search results, which is notable for a platform that claims "600+ leading brands" as clients.

Findings

1. No llms.txt file published High

   The site returns a 404 for /llms.txt, meaning there is no AI-friendly content map to help crawlers discover key pages like Helix AI.

   What to change: Publish an llms.txt file listing the most important pages for AI consumption, especially /helix-ai and /helix.

2. Robots.txt lacks AI-bot-specific rules Medium

   The robots.txt uses a single User-agent: * rule with no AI-bot-specific directives, missing an opportunity to signal preferred crawling behavior for GPTBot, ClaudeBot, and others.

   What to change: Add explicit rules for AI bots (e.g., GPTBot, ClaudeBot) to allow full access and optionally point to the llms.txt.

3. LLM cold knowledge is 2-3 years behind current products High

   LLMs describe DeepIntent as a Propel Media acquisition with products 'Audience Engine' and 'DeepIntent Outcomes', but the site shows an independent company with Cortex, Helix, and Helix AI, and no mention of Audience Engine.

   What to change: Publish an llms.txt and ensure key product pages are well-indexed and contain clear, up-to-date descriptions to correct LLM knowledge.

4. BreadcrumbList schema absent from most pages Medium

   Only the blog article includes BreadcrumbList schema; all other pages lack it, reducing navigational context for AI crawlers.

   What to change: Add BreadcrumbList structured data to all main pages to improve AI understanding of site hierarchy.

5. Core DSP platform lacks Product schema Medium

   The platform page uses SoftwareApplication schema but not Product schema, missing applicationSubCategory and operatingSystem details that would help AI understand the offering.

   What to change: Add Product schema with applicationSubCategory and operatingSystem to the platform page.

6. Footer renders unrendered Webflow template variable Low

   The footer displays '© {{year}} DeepIntent' as literal text, indicating a Webflow template variable that was not compiled, which appears unprofessional to crawlers.

   What to change: Replace the template variable with a static year (e.g., 2026) in the footer.

7. Organization schema has empty sameAs array Low

   The homepage Organization schema includes a sameAs array with no URLs, even though social media links exist in the footer, missing an opportunity to link to external profiles.

   What to change: Populate the sameAs array with actual social media profile URLs (e.g., LinkedIn, Twitter).

8. No external review site presence detected Medium

   Web searches for reviews on G2, TrustRadius, or similar platforms returned no results, which is notable for a platform claiming 600+ clients.

   What to change: Encourage clients to leave reviews on G2 or TrustRadius and link to those profiles from the site.

What's working

• All pages return full server-rendered HTML — Every page delivers complete HTML content without JavaScript dependency, ensuring AI crawlers can read all text content directly.
• All major AI bots allowed with full content — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and others receive 200 responses with full HTML, and no AI bot is blocked except Bytespider.
• Consistent JSON-LD structured data across pages — All major pages include appropriate schema types (Organization, SoftwareApplication, FAQPage, ItemList, Person), providing rich context for AI crawlers.
• Frequently updated newsroom and blog — The news page and DeepDive blog are updated multiple times per week with press releases and articles, signaling an active company to AI crawlers.
• FAQPage schema on Helix page with answers — The /helix page includes FAQPage structured data with three questions and answers, plus a Review schema, helping AI extract key product details.
• SoftwareApplication schema on Helix AI page — The /helix-ai page uses SoftwareApplication schema with PreOrder availability, clearly marking the product as an AI tool.
• Leadership page with detailed profiles and LinkedIn links — The /leadership page lists 14 team members with headshots and LinkedIn URLs, providing authoritative person data for AI crawlers.