Skip to main content
Kimi (Moonshot AI’s assistant) ran a textbook AI-visibility content play — and its results show that being retrieved is not the same as being recommended. David Konitzny’s case study (full article, Jul 2026):
  • The play: mass uploads of listicles and how-to articles on kimi.com — 882 pages live on a single day (June 6, 2026; 52% how-tos, 27% listicles), two more bulk uploads around the Kimi 3 release (July 16, 2026), all translated into six-plus languages. Kimi ranked itself #1 in its own “Top 10 AI App Builders” listicle.
  • Retrieval worked fast: within weeks Kimi ranked top-5 among retrieved sources across matched prompts with an 11.2% retrieval share, driven almost exclusively by the new pages. Retrieval was strongest in Italy and Germany (multilingual entry points), just 0.8% in the US.
  • Brand visibility lagged: Kimi ranked only sixth in actual brand mentions at 5% visibility share, versus Claude’s 15%. Model differences were large — Gemini mentioned Kimi at 3%, ChatGPT at just 0.4%.
The takeaway: large-scale, AI-focused content can enter LLM retrieval pipelines within weeks, but retrieval visibility does not automatically convert into brand recommendations — track both separately. Metric definitions: Retrieval vs citation. Source: Is Kimi using an AI visibility content strategy?, David Konitzny, LinkedIn, Jul 2026.