Anthropic · last updated 2026-05-21
Claude
Anthropic's Claude is a conversational AI with optional web search grounding (introduced in 2025). Claude leans more cautious about citation than ChatGPT or Perplexity. It grounds when the model judges fresh information is necessary and surfaces sources in a side panel rather than inline. The grounding behavior is more conservative, the citations more curated, and the answer style more long-form.
Grounding model
On-demand web search invoked when the model decides recency or external information is needed. Anthropic's own crawler (`ClaudeBot`) is the primary indexer. Citations are surfaced as a side panel of source cards.
Citation style
Side-panel source cards rather than inline footnotes. Each card shows title, domain, and snippet. Tap-through opens the source in a new tab. The citation surface is less prominent than Perplexity but cleaner than older chat UIs.
What it rewards
- Long-form, well-structured content the model can ground complex multi-part questions against
- Explicit authorship, dating, and methodology (Claude is more cautious about citing low-trust sources)
- Self-contained explanatory writing that survives extraction into Claude's longer-form answer style
- Schema.org Article and Person markup for E-E-A-T signals
What it penalizes
- Thin, listicle-heavy content with weak per-item analysis
- Pages with obvious affiliate or sponsored content that Anthropic's filtering tags as low-trust
- Robots.txt blocking ClaudeBot (the crawler will skip you entirely)