Claude

Claude is Anthropic's assistant. When a question needs current information it answers with web search, citing the sources it used. Lokrix runs Claude live by default, with web search enabled so it reflects how people actually get product answers from it.

How it grounds

Claude grounds via web search. It tends to search deliberately — invoking it when the question is time-sensitive or benefits from external sources — and to prefer clear, authoritative pages it can quote and attribute. Being findable on the open web for a question, and giving Claude something concrete to cite, is what gets you surfaced.

How each answer engine grounds its answersSeven answer engines on the left connect to their grounding sources on the right: ChatGPT to the Bing index; Google AI Overviews, AI Mode and Gemini to the Google index and Search; Perplexity, Grok and Claude to live web search.ChatGPTGoogle AI OverviewsGoogle AI ModePerplexityGeminiGrok (xAI)ClaudeBing indexGoogle index / SearchLive web search

What moves your score

  • Quotable, well-sourced content: statistics, direct quotations and cited sources give Claude concrete material to attribute — the top measured on-page lever (Princeton GEO, KDD 2024).
  • Clarity and structure: pages that answer the question directly, with clean headings, are easier to extract from.
  • Web discoverability for the buyer questions in your prompt universe.
  • Earned-media corroboration (Reddit, Wikipedia, G2) that reinforces your authority.

Each Claude run is sampled K times into a presence probability with a 95% confidence interval — never a single-run yes/no; see sampling & confidence and how Claude contributes to the per-engine breakdown.