Joshua Opolko

The GEO Observatory

A live dashboard of which AI and search engines crawl this site, which pages they pull for answers, and what they ignore. Parsed from raw server logs, verified against operator IP ranges. Rolling 8-day window (2026-07-05 to 2026-07-12). Generated 2026-07-12 06:40 UTC.

AI engines fetch a page before they cite it, so this is the citation pipeline in the open. The AI live-fetch count is pages being pulled into answers right now; the blind spots are pages AI never sees; the Verified checks separate real bots from impostors. This is the "watch your logs" step from the GEO Field Manual, made live.

Who's actually visiting

Real IP analysis: ip-api.com org lookup on non-bot request IPs, classified by network type. 42,913 unique IPs in the past 7 days. Updated weekly (Sundays) · data as of 2026-07-05.

Rank-tracking / residential proxy (M247 Ltd Paris, FR Onetsolutions, 07 m247 3)
1%
Cloud / VPS scraper (Microsoft Azure Cloud (brazilsouth), Microsoft Azure Cloud (japanwest), Microsoft Azure Cloud (japaneast))
25%
Real visitor (residential ISP) (Sympatico HSE, Comcast Cable Communications, LLC, Spectrum)
2%
Unclassified (Techoff SRV Limited, Privex Sweden, Bharti Airtel Limited)
72%

• Techoff SRV Limited (4 nodes, 14,367 hits): Andorra/Netherlands-based infrastructure, single-visit behavior across unknown pages — consistent with automated content harvester or mirror scraper operating at scale.

• (2a07:e01:3:1f8::1) Privex Sweden, 7,471 hits: IPv6 residential proxy from Sweden, single visit to unknown pages — rank-tracking proxy inflating impressions without genuine user engagement.

• Microsoft Azure Cloud (7 nodes, 2,972 hits): Brazil, Hong Kong, Singapore, India, Spain, Japan — distributed cloud scraper with single-visit pattern, likely commercial content indexing or SEO aggregation service.

• (129.150.38.224) Oracle Corporation Singapore, 513 hits: Single-visit behavior suggests Oracle-hosted bot or crawler, possibly legitimate infrastructure audit but operating as automated scraper.

• AWS EC2 (2 nodes, 674 hits): Australia/France regions, single-visit pattern — commodity cloud hosting used for distributed scraping, likely content replication or price-comparison crawler.

• Cross-reference with GSC: /claude-fable-5-issues-fixes/ shows 4,753 impressions but only 58 clicks (1.2% CTR) — rank-tracking and proxy traffic are heavily inflating this site's GSC impression count relative to real search interest; /agent-zero/ and /ollama/ show similar patterns (low CTR despite impressions), suggesting proxy bots are triggering SERP appearances without actual user demand.

Real human traffic represents approximately 2% of total unique IPs this site received, concentrated in /claude-fable-5-issues-fixes/ and /d-lusion-rubberduck/ (35.3% CTR) where legitimate clicks actually occur.

Search Console: top pages by impression (last 7 days)

PageImpressionsClicksCTRAvg pos
/claude-fable-5-issues-fixes/4,753581.2%8.1
/agent-zero/1,37330.2%13.0
/ollama/1,23630.2%35.8
/searxng-self-hosted-guide/59130.5%8.4
/claude-code-specification-workflow-mcp/36300%11.3
/dify-self-hosted-guide/17300%26.2
/geo-ai-citation/12600%37.7
/perplexica-self-hosted-guide/11010.9%17.9
/geo-field-manual/3700%15.9
/cyubevr/2700%9.2
/babylon-js-clouds/1815.6%13.1
/d-lusion-rubberduck/17635.3%4.9
/claude-seo/1200%16.9
/hometurf/900%6.4
/anythingllm-self-hosted-guide/600%30.5

Red CTR = impression volume likely inflated by rank trackers (they query Google, not real searchers).

3,848
AI crawler hits
176 pages fetched
2,469
AI live-fetches
pulled into answers now
1,470
Search-engine hits
classic indexing
3,906
Verified bots
3906/4343 IP-checked
3,062
Scanner probes
hostile, all 404
Updated 2026-07-12 06:40 UTC · data 2026-07-05 → 2026-07-12 · rebuilt daily Dates are first-seen in the server log, not publication dates.

AI crawler activity, last 14 days

Daily AI hits, stacked by what the crawler is doing. Hover any day for the breakdown.

How to read this: daily counts of AI-crawler hits from 2026-07-05 to 2026-07-12, one bar per day, each segmented by crawler intent (AI live-fetch, AI search, AI training). Busiest day was 2026-07-07 with 774 AI hits. Data window first-seen 2026-07-05, latest 2026-07-12; last updated 2026-07-12 06:40 UTC.

What the AI crawlers want

Every AI fetch in the window, split by intent.

How to read this: the 3,848 AI fetches in this window split by intent: 295 ai search, 1,084 ai training, 2,469 ai live-fetch. Window 2026-07-05 to 2026-07-12.

Top AI engines

By hits. Click one to filter the page explorer below.

How to read this: AI crawlers ranked by hit count, 2026-07-05 to 2026-07-12. Top engine ChatGPT user-fetch with 2,435 fetches, then Meta AI (285).

Live Desk · July 12 edition

AI assistants pulled 59 pages of this site mid-answer in the window (2,469 live fetches in all). The most active was LiteLLM Proxy: One OpenAI-Compatible API for 100-Plus LLM Providers, fetched 462× by ChatGPT: live-answer agents, meaning it was read to compose a reply to a real question, not archived for training.

Pages AI is fetching

Search, filter by crawler intent or engine, and click any row to see which bots pull it. Showing the top 25 by default. 176 pages have AI activity.

AllLive-fetchAI searchAI training
PageAI hitsLive-fetchTop type
/litellm-proxy-guide/470462AI live-fetch
/470215AI live-fetch
/dify-self-hosted-guide/418401AI live-fetch
/searxng-self-hosted-guide/322305AI live-fetch
/agent-zero/235225AI live-fetch
/n8n-self-hosted-guide/211200AI live-fetch
/anythingllm-self-hosted-guide/153146AI live-fetch
/claude-fable-5-issues-fixes/138129AI live-fetch
/crewai-setup-production-guide/6153AI live-fetch
/claude-code-specification-workflow-mcp/4842AI live-fetch
/ollama/4133AI live-fetch
/librechat-self-hosted-guide/3222AI live-fetch
/open-webui-self-hosted-guide/3118AI live-fetch
/vision-dominates-half-your-brains-processing-power/3123AI live-fetch
/medical-training/2927AI live-fetch
/hometurf/2916AI live-fetch
/perplexica-self-hosted-guide/2716AI live-fetch
/psychedelic-vr-visual-effects-meta-quest/251AI training
/kidsevents/2314AI live-fetch
/qwen2-5-vl/2317AI live-fetch
/driftlights/190AI training
/geo-observatory/190AI training
/cyubevr/1810AI live-fetch
/claude-seo/180AI training
/aiscan/162AI training

Blind spots

Indexed pages AI has not fetched in this window, your fix-list (orphan pages, weak internal links, stale content). 0 total.

Freshness: how fast AI picks up updates

PageUpdatedFirst AI fetchLag
/kidsevents/neighbourhoods/2026-07-062026-07-060d
/d-lusion-rubberduck/2026-07-062026-07-093d

AI referrals: humans who clicked through

47 click-throughs from AI platforms — Copilot/Bing leads with 32. Last 7 days: 22 (18 Copilot/Bing, 2 Copilot, 1 Claude.ai, 1 Perplexity). These are real visitors whose browser sent a Referer header from an AI platform — proof the citation converted to a visit. Country and ISP come from the real visitor IP (via Cloudflare).

DatePlatformPageLocationISP / Org
2026-07-12Copilot/ollama/🇺🇸 Maiden, United StatesSpectrum
2026-07-11Copilot/Bing/searxng-self-hosted-guide/ Foshan, ChinaChinanet GD
2026-07-11Copilot/Bing/🇺🇸 New York, United StatesLogicWeb Inc
2026-07-10Copilot/Bing/litellm-proxy-guide/🇮🇳 Chennai, IndiaNetskope MAA
2026-07-10Copilot/Bing/geo-ai-citation/🇺🇸 Portland, United StatesCloudflare, Inc.
2026-07-10Copilot/Bing/litellm-proxy-guide/🇩🇪 Fuldabrück, GermanyVodafone NRW GmbH
2026-07-10Copilot/Bing/kidsevents/🇨🇦 Toronto, CanadaThe Hospital Sick Children
2026-07-10Copilot/Bing/kidsevents/🇨🇦 Toronto, CanadaRogers Communications Canada Inc
2026-07-09Copilot/Bing/litellm-proxy-guide/ Guangzhou, ChinaChina Mobile
2026-07-09Copilot/Bing/searxng-self-hosted-guide/🇩🇪 Berlin, Germanyecotel communication ag
2026-07-09Copilot/Bing/crewai-setup-production-guide/ Guangzhou, ChinaChina Unicom
2026-07-09Copilot/Bing/crewai-setup-production-guide/ Jinrongjie, ChinaChina Unicom Network
2026-07-09Copilot/Bing/litellm-proxy-guide/ Beijing, ChinaCHINANET
2026-07-09Copilot/Bing/d-lusion-rubberduck/🇳🇱 Heerenveen, The NetherlandsVodafone Ziggo
2026-07-08Copilot/vision-dominates-half-your-brains-processing-powe🇳🇱 Amsterdam, The NetherlandsHotel StadsThuis B.V.
2026-07-07Claude.ai/claude-fable-5-issues-fixes/🇯🇵 Chiyoda City, JapanSOFTBANK Corp
2026-07-07Perplexity/perplexica-self-hosted-guide/🇩🇪 Frankfurt am Main, GermanySERV.HOST GROUP LTD
2026-07-07Copilot/Bing/dify-self-hosted-guide/ Jinrongjie, ChinaChina Mobile
2026-07-07Copilot/Bing/kidsevents/🇺🇸 Chicago, United StatesZscaler, Inc
2026-07-07Copilot/Bing/kidsevents/🇨🇦 Toronto, CanadaTELUS Mobility-Ontario
2026-07-06Copilot/Bing/🇺🇸 Ashburn, United StatesBraveWay
2026-07-06Copilot/Bing/🇨🇦 Toronto, Canada3xK Tech GmbH
2026-07-04Copilot/Bing/searxng-self-hosted-guide/🇵🇭 Tagaytay, PhilippinesEastern Telecoms Philippines, Inc.
2026-07-04Copilot/Bing/dify-self-hosted-guide🇺🇸 Ashburn, United StatesAWS EC2 (us-east-1)
2026-07-04Copilot/Bing/dify-self-hosted-guide/🇺🇸 Ashburn, United StatesAWS EC2 (us-east-1)
2026-07-04Claude.ai/searxng-self-hosted-guide/🇺🇸 Boston, United StatesCdnext BOS
2026-07-04Copilot/Bing/kidsevents/🇨🇦 North York, CanadaRogers Communications Canada Inc
2026-07-04Copilot/Bing/kidsevents/🇨🇦 Richmond Hill, CanadaRogers Cable Inc. WLFDLE
2026-07-03Copilot/models/🇮🇳 Hyderabad, IndiaReliance Jio Infocomm Limited
2026-07-03Perplexity/perplexica-self-hosted-guide/🇺🇸 Portland, United StatesCloudflare, Inc.
2026-07-02Copilot/Bing/dify-self-hosted-guide/ Guangzhou, ChinaChina Mobile
2026-07-02Copilot/vision-dominates-half-your-brains-processing-powe🇳🇱 Maastricht, The NetherlandsUniversiteit Maastricht
2026-07-02Copilot/Bing/geo-field-manual🇺🇸 Ashburn, United StatesAWS EC2 (us-east-1)
2026-07-02Copilot/Bing/geo-field-manual/🇺🇸 Ashburn, United StatesAWS EC2 (us-east-1)
2026-07-02Gemini/geo-field-manual🇦🇪 Fujairah, United Arab EmiratesEtisalat
2026-07-02Gemini/geo-field-manual/🇦🇪 Fujairah, United Arab EmiratesEtisalat
2026-07-01ChatGPT/searxng-self-hosted-guide/🇺🇸 Portland, United StatesCloudflare, Inc.
2026-07-01Copilot/Bing/🇺🇸 Ashburn, United States3xK Tech GmbH
2026-07-01Copilot/Bing/vision-dominates-half-your-brains-processing-powe Hanoi, VietnamVietnam Internet Network Information Cen
2026-07-01Copilot/architectural-visualization/🇪🇸 Valencia, SpainXtra Telecom S.A.
2026-07-01Perplexity/kidsevents/🇨🇦 Mississauga, CanadaRogers Communications Canada Inc
2026-07-01Perplexity/perplexica-self-hosted-guide/🇺🇸 Sacramento, United StatesAT&T Internet Services
2026-06-30Claude.ai/searxng-self-hosted-guide/🇺🇸 Portland, United StatesCloudflare, Inc.
2026-06-30Copilot/Bing/🇳🇱 Amsterdam, The NetherlandsTechoff SRV Limited
2026-06-30Copilot/Bing/vision-dominates-half-your-brains-processing-powe Hanoi, VietnamVietnam Internet Network Information Cen
2026-06-30Copilot/Bing/kidsevents/🇨🇦 Toronto, CanadaSympatico HSE
2026-06-29Copilot/Bing/ Andorra la Vella, AndorraTechoff SRV Limited

Who is crawling

Search engines

Classic index crawlers, by hits.

Bingbot900
Googlebot180
Petal (Huawei)179
Yandex113
Applebot46
DuckDuckGo37
Seznam15

Are they who they claim?

Crawler traffic checked against each operator's published IP ranges and reverse-DNS.

Full crawler tables (raw numbers)

AI crawlers

CrawlerTypeHitsVerifiedLast
ChatGPT user-fetchAI live-fetch2,4352413/24352026-07-12
Meta AIAI training285pending2026-07-12
AmazonbotAI training279pending2026-07-12
Anthropic ClaudeBotAI training1800/1802026-07-12
ByteDance BytespiderAI training178pending2026-07-12
PerplexityAI search159138/1592026-07-12
ChatGPT searchAI search136104/1362026-07-12
OpenAI GPTBotAI training8340/832026-07-12
Claude user-fetchAI live-fetch290/292026-07-11
CohereAI training19pending2026-07-07
Google-ExtendedAI training180/182026-07-07
Common CrawlAI training17pending2026-07-09
Applebot-ExtendedAI training150/152026-07-07
Anthropic (other)AI training70/72026-07-07
Perplexity user-fetchAI live-fetch50/52026-07-07
You.comAI training3pending2026-07-09

Search engines

CrawlerTypeHitsVerifiedLast
BingbotSearch index900885/9002026-07-12
GooglebotSearch index180174/1802026-07-12
Petal (Huawei)Search index179pending2026-07-12
YandexSearch index113113/113 ✓2026-07-12
ApplebotSearch index4639/462026-07-12
DuckDuckGoSearch index370/372026-07-09
SeznamSearch index15pending2026-07-12

Under attack: what scanners hunt for

3,062 requests probing for exposed files, admin panels, and known exploits, every one a 404 here. 114 forged a real crawler's user-agent to blend in.

Bot traffic reality

Every bot request in the window. Hostile scanning dwarfs the legitimate crawlers.

Of 8,380 bot requests: 3,062 scanner probes (37%), 3,848 AI, 1,470 search.

What they hunt for

Probes grouped by target. Mostly WordPress, on a site that runs none.

By target: 2,414 wordpress, 73 secrets & config, 49 admin / rce, 233 git exposure, 293 other.

Top probed paths

The most-hammered URLs. None of them exist here.

/wp-login.php632
/aiscan/api/scan/stream80
/wp-content/plugins/hellopress/wp_filemanager.php61
/aiscan/api/report/c716161291e141
/wp-content/uploads/23
//wp-includes/css/dist/23
/wp-content/themes/index.php21
/wp-content/index.php19
//wp-includes/assets/19
/wp-content/admin.php18
/wp-content/uploads/2026/06/claude-seo-featured.png18
/.git/config18
Full probed-path table
Probed pathHits
/wp-login.php632
/aiscan/api/scan/stream80
/wp-content/plugins/hellopress/wp_filemanager.php61
/aiscan/api/report/c716161291e141
/wp-content/uploads/23
//wp-includes/css/dist/23
/wp-content/themes/index.php21
/wp-content/index.php19
//wp-includes/assets/19
/wp-content/admin.php18

Verified counts cover traffic since real-IP logging was enabled, filling in daily. Log retention is ~2 weeks, so "blind spots" means "not fetched recently." The playbook: GEO Field Manual, and the answers it produces: GEO answers.