kx.com — AI Site Grade

Analysis

KX's AI-visibility posture is excellent on the infrastructure layer but undermined by a product-naming gap that leaves cold LLM knowledge stuck describing a legacy product line that no longer matches the site's own messaging.

Crawler Access

Every major AI crawler — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, ChatGPT-User, anthropic-ai, Applebot-Extended — receives a full 200 response with the same 237KB HTML payload as a browser. The robots.txt explicitly allows GPTBot, Claude-Web, Perplexity-Bot, and others while blocking Bytespider, Deepseek, MistralAI, and YouBot. The site runs on nginx on Azure (A record: 3.212.210.244) with no WAF or CDN layer that discriminates by UA. DNS TXT records confirm proactive AI-verification handshakes: openai-domain-verification, anthropic-domain-verification, and apple-domain-verification are all present. /llms.txt returns a 404 — a missed opportunity for a site with this much structured technical content.

Cold-Knowledge Gap

Asked cold, a frontier LLM describes KX as "the kdb+ time-series database" company whose core product uses the "proprietary q programming language," with "KX Insights" as the cloud platform. The site itself has completely re-centered its narrative around KDB-X (GA announced April 2026), a unified compute engine for time-series, vector search, and GPU-accelerated AI. The homepage's H1 is "World's Fastest Analytics for Time-Series and Vector Data." The cold model knows nothing about KDB-X, GPU acceleration, the NVIDIA partnership, or the vector-database positioning. The gap is not subtle: the model describes a 2023-era product line while the site is aggressively selling a 2026-era platform.

Schema Posture

The homepage carries Organization and WebSite JSON-LD with SearchAction, but no SoftwareApplication schema for KDB-X, KDB.AI, or any product. The KDB-X product page (/products/kdb-x/) and the KDB.AI page (/products/kdb-ai/) have zero JSON-LD — no schema at all. The glossary page ("What is a Vector Database?") does have Article schema with proper datePublished/dateModified. Given that the site explicitly positions itself as a vector database and analytics platform, the absence of SoftwareApplication + SoftwareApplicationCategory on product pages is a structural weakness for AI-driven answer engines that extract structured product data.

Content Signals

The homepage is text-rich (~1,600 words visible) with strong comparison language ("KDB-X vs QuestDB, ClickHouse, TimescaleDB and InfluxDB"), benchmark claims (STAC-M3, TSBS), and a pricing table on the KDB-X page. No FAQ schema is used anywhere despite the glossary functioning as a de facto FAQ. The blog contains 50+ posts with solid technical depth. A notable anomaly: the homepage JSON-LD lists dateModified: 2026-05-01 — a future date — suggesting a WordPress plugin misconfiguration or a pre-scheduled update that leaked into the schema.

External Signals

Web search returns minimal third-party coverage of KDB-X specifically. The brand's external footprint (customer stories from RBC Capital Markets, ADSS, Axi, BWT Alpine F1 Team) is strong but the new product narrative has not propagated into the general web corpus yet. The G2 seller profile is linked in the homepage schema but no recent review data surfaced in search.

Findings

1. Cold LLM knowledge describes legacy kdb+ product line, not current KDB-X platform High

   A frontier LLM queried about KX describes the company as a kdb+ time-series database vendor using the proprietary q language, with KX Insights as the cloud platform. The site has re-centered its narrative around KDB-X, a unified compute engine for time-series, vector search, and GPU-accelerated AI, announced GA in April 2026. The cold model knows nothing about KDB-X, GPU acceleration, the NVIDIA partnership, or the vector-database positioning.

   What to change: Publish an /llms.txt file that explicitly describes KDB-X, KDB.AI, and the current product positioning. Ensure the homepage and product pages contain clear, crawlable text that reinforces the new narrative.

2. /llms.txt returns 404, missing opportunity to guide AI crawlers Medium

   The site does not serve an /llms.txt file, which is a standard way to provide structured guidance to AI crawlers about which pages to use for training and retrieval. Given the site's technical depth and product documentation, this is a missed opportunity to shape AI knowledge.

   What to change: Create an /llms.txt file that lists key pages such as product pages, blog posts, and glossary entries, with brief descriptions to help AI crawlers prioritize content.

3. Product pages lack SoftwareApplication schema High

   The KDB-X product page (/products/kdb-x/) and the KDB.AI page (/products/kdb-ai/) have zero JSON-LD schema. The homepage has Organization and WebSite schema but no SoftwareApplication markup for any product. This limits the ability of AI-driven answer engines to extract structured product data.

   What to change: Add SoftwareApplication JSON-LD schema to all product pages, including properties like name, description, applicationCategory, operatingSystem, and offers.

4. Homepage JSON-LD lists future dateModified (2026-05-01) Low

   The homepage's JSON-LD schema contains a dateModified value of 2026-05-01, which is a future date at the time of audit. This may confuse crawlers and reduce trust in the schema data.

   What to change: Correct the dateModified in the homepage JSON-LD to reflect the actual last modification date, or remove it if not accurately maintained.

5. No FAQ schema used despite glossary functioning as FAQ Medium

   The site has a glossary section (e.g., /glossary/what-is-a-vector-database/) that answers common questions, but no FAQPage schema is implemented. This reduces the chance of appearing in rich results for question-based queries.

   What to change: Add FAQPage schema to glossary pages that contain question-answer pairs, or create a dedicated FAQ page with schema markup.

6. Minimal third-party web coverage of KDB-X Medium

   Web searches for KDB-X and related terms return zero results, indicating that the new product narrative has not propagated into the general web corpus. This limits external signals that AI models can use to validate the site's claims.

   What to change: Increase PR and content marketing efforts to generate third-party articles, reviews, and case studies about KDB-X. Encourage customers to publish reviews on platforms like G2.

7. No recent G2 reviews surfaced for KX products Low

   A web search for KX kdb+ reviews on G2 returned zero results, despite the homepage schema linking to a G2 seller profile. This suggests limited social proof on a key review platform.

   What to change: Encourage customers to leave reviews on G2 and other review platforms to build social proof and external signals.

What's working

• All major AI crawlers allowed and served full HTML — Every major AI crawler (GPTBot, ClaudeBot, PerplexityBot, Google-Extended, etc.) receives a 200 response with the same 237KB HTML payload as a browser. The robots.txt explicitly allows these bots while blocking only problematic ones like Bytespider and Deepseek.
• DNS TXT records confirm AI-verification handshakes — The domain has openai-domain-verification, anthropic-domain-verification, and apple-domain-verification TXT records, proactively verifying ownership to AI providers.
• Homepage is text-rich with strong comparison and benchmark data — The homepage contains ~1,600 words of visible text, including comparisons against QuestDB, ClickHouse, TimescaleDB, and InfluxDB, as well as benchmark claims (STAC-M3, TSBS). This provides substantial content for AI crawlers to index.
• Glossary page uses Article schema with proper dates — The glossary page 'What is a Vector Database?' includes Article JSON-LD with datePublished and dateModified, which helps search engines understand the content freshness.
• Homepage has Organization and WebSite schema with SearchAction — The homepage includes Organization and WebSite JSON-LD with SearchAction, providing basic structured data about the company and site search capability.
• Infrastructure uses nginx on Azure with no UA-based blocking — The site runs on nginx on Azure with no WAF or CDN layer that discriminates by user agent, ensuring consistent access for all crawlers.
• Sitemap contains 80 URLs and is properly indexed — The sitemap at /sitemap.xml returns 200 and lists 80 URLs, providing a clear map of the site's content for crawlers.