New Profound research analyzing 24,135 Claude and Claude Code responses (plus the top 1,000 webpages each agent visited over 30 days) finds the two products behave like separate Answer Engines despite sharing the same underlying models. Claude Code searches the web in just 13% of responses versus Claude's 93%; even though both mention a similar number of brands per response (6.6 vs 5.2), those brands overlap only 20% on average — far less than the ~50% overlap between two separate Claude responses to the same prompt.
The two agents also read entirely different parts of the web: nearly three-quarters of Claude Code's observed page visits went to documentation, pricing, and informational pages, versus just 5% for Claude; conversely, 60% of Claude's visits went to robots.txt files, sitemaps, and homepages, versus only 4% for Claude Code. Claude Code's responses are also shorter (322 vs 459 words) but far more structured (94% include a list, 54% a table, vs. 56%/11% for Claude).
Profound's conclusion: harnesses — not just the underlying model — materially change citation and mention behavior, so brands need to measure and optimize for each harness separately rather than treating "Claude" as one Answer Engine.
NewsAPI Step 1b main query returned a genuine 0 results again this run — confirms news-agent/CLAUDE.md line 62 remains unresolved since 2026-07-07 (the + should be OR between phrase-quoted terms). The competitor query (1c) returned 8 raw hits, 0 genuine (a celebrity obituary syndicated twice, football/sport trivia, wildlife explainers, a political op-ed, and a Nature piece on AI consciousness — no GEO-competitor signal at all today).
Otterly.ai's blog is unreachable again — connection refused on today's fetch, one day after being reachable following the Aug 22–23 outage. This is now a recurring, not isolated, reliability problem with Otterly's blog server; worth a direct status check outside the normal sweep if a third instance occurs this week.
shopify.dev's changelog feed.xml still returns HTTP 500 — direct page-render fallback caught a genuinely new item (script tags deprecation) that the broken feed would otherwise have hidden; the fallback continues to work reliably.
Anthropic Newsroom RSS (rss.xml) still 404s — direct fetch of the newsroom page confirms no post beyond Aug 14's watermark FAQ, so no signal was lost.
Firecrawl: 3/3 scrapes attempted; 2 of 3 (both TechCrunch pieces — Hugging Face and OpenAI's agent feature) returned a Cloudflare Turnstile challenge banner prepended to the markdown — same recurring failure mode as recent runs, but full article text followed cleanly beneath the boilerplate in both cases. The Profound scrape came back fully clean.