armature ai traffic

Be in the next model's priors.

You need to be in the training sets, the AI search indexes, and the answers. Armature shows when OpenAI, Anthropic, Google, Perplexity, and other AI systems actually fetch your content.

Start tracking AI traffic → completely free View a sample dashboard
ChatGPT Claude Claude Code Gemini Perplexity Cursor GPTBOT · CCBOT +50 yoursite.com /docs/quickstart /pricing /blog/launch /docs/api /llms.txt AI ANSWERS USER-TRIGGERED SEARCH INDEXES DISCOVERY MODEL TRAINING DATA COLLECTION

No browser JavaScript needed. Your human analytics stays as it is.

the stakes

Agents decide in steps: you must be in each of them.

A coding agent recalls its priors , searches the web , reads what it finds , then picks. Miss a step and you are out of the run.

01 · its priors

It starts from what the model trained on.

→ be in the training data

model training

02 · web search

It queries the AI indexes.

→ be in them

search indexes

03 · it reads

It fetches the pages it found.

→ be readable

ai answers

04 · the pick

It chooses from what it saw.

→ you, or your competitors

Armature tracks the crawls behind every step. You see exactly where you exist for agents, and where you do not.

the dashboard

Know who reads what, and why.

Every crawl attributed to a system, a bot, a page, and a purpose. Filter by website and time range, and watch the traffic move.

site: yoursite.com ▾ range: last 30 days ▾ Purpose AI company Bot sample data
Total AI crawls 12,384 AI traffic only
AI answer fetches 3,164 User-triggered fetches
Search crawls 6,908 Indexing and discovery
Training crawls 2,312 Model data collection
Verified traffic 87% IP-verified requests
Traffic over time · daily crawls
AI answers Search indexes Training crawlers
0 150 300 Jul 29 Aug 7 Aug 17 Aug 27 search answers training

Top crawlers

identities with the most requests
OAI-SearchBotOpenAI2,014
GooglebotGoogle1,564
ChatGPT-UserOpenAI1,318
PerplexityBotPerplexity1,312
Claude-SearchBotAnthropic1,242
GPTBotOpenAI1,096

Top pages

ranked by requests
/docs/quickstart2,981
/2,140
/docs/api/auth1,655
/blog/mcp-guide1,209
/pricing884
/llms.txt733

Recent crawler activity

live sample
WhenCrawlerPurposePageConfidence
just nowChatGPT-UserAI answer/docs/quickstartVerified
14s agoPerplexityBotSearch/pricingVerified
31s agoGPTBotTraining/blog/launch-weekVerified
48s agoClaude-UserAI answer/docs/api/authVerified
1m agoOAI-SearchBotSearch/llms.txtVerified
2m agometa-externalagentTraining/docs/self-hostingUA only

Sample data for illustration. Your dashboard shows live traffic from your own domains.

setup

One prompt. Then the traffic appears.

Create a key and paste one prompt into your coding agent. It installs Armature and sends a production test.

set up Armature AI Traffic on yoursite.com ⏎

Your coding agent does the installing. You watch the first event land.

the directory

The bots behind the traffic.

Armature classifies 52 known AI crawler identities by who runs them and why they fetch. These are the names you will see in your dashboard.

AI answers

15 bots · user-triggered
ChatGPT-UserOpenAI Fetches a page while ChatGPT answers a user.
Claude-UserAnthropic Fetches a page while Claude answers a user.
Perplexity-UserPerplexity Fetches sources for Perplexity answers and citations.
CopilotMicrosoft Fetches pages for Microsoft Copilot answers.
Qwen-UserAmzn-UserDuckAssistBotGoogle-AgentGoogle-NotebookLMGoogle-Read-AloudGoogleAgentmeta-externalfetcherMistralAI-UserKimi-UserGrok-DeepSearch

Search indexes

15 bots · discovery
OAI-SearchBotOpenAI Crawls for the index behind ChatGPT search.
Claude-SearchBotAnthropic Discovers and refreshes pages for Claude search.
PerplexityBotPerplexity Keeps the Perplexity index fresh.
GooglebotGoogle The Search index, which also feeds AI Overviews.
Amzn-SearchBotApplebotBaiduspiderTikTokSpiderGoogle-InspectionToolBingbotmsnbotMistralAI-IndexKimi-SearchBotYouBotxAI-SearchBot

Training crawlers

13 bots · model data
GPTBotOpenAI Collects public pages that may improve future OpenAI models.
ClaudeBotAnthropic Collects public pages that may improve future Claude models.
CCBotCommon Crawl Builds the open crawl datasets many labs train on.
Google-CloudVertexBotGoogle Site crawls used for building Vertex AI agents.
QwenBotAI2BotERNIEBotBytespidercohere-aicohere-training-data-crawlerDeepSeekBotmeta-externalagentKimiBot

Other AI bots

9 bots · everything else

Ad fetchers, link previews, and bots whose purpose is not pinned down yet.

AmazonbotDoubaobotGoogleOtherFacebookBotfacebookexternalhitOAI-AdsBotGrokBotxAI-BotxAI-Web-Crawler

New bots join the catalog as they appear. Bot-like agents we cannot name yet show up as not identified.

why they come

Three reasons AI systems fetch your content.

Every crawl has a purpose. Armature classifies each request, so you can tell an answer fetch from an index crawl from a training run.

01 · AI answers

See when an assistant fetches one of your pages to answer a user.

ChatGPT, Claude, and Perplexity fetch pages on demand while a user waits. Each fetch means an assistant opened your page while answering a real question.

Logged as it happens and attributed to the assistant that fetched.

02 · Search indexes

Know which pages AI search crawlers discover and refresh.

OAI-SearchBot, Claude-SearchBot, and PerplexityBot crawl to build the indexes behind AI search. Watch new pages get discovered and old ones get refreshed.

A crawl shows discovery and refresh. It does not guarantee a ranking.

03 · Model training

See when training crawlers request content that may help shape future models.

GPTBot, ClaudeBot, and CCBot collect public pages for model development. Know when they come, what they take, and how often they return.

A request does not guarantee inclusion in a training dataset.

what you can answer

The questions it answers.

01Are AI systems finding my new content?time filter · top pages
02Which AI labs read my website?ai company view
03Which pages matter most to them?top pages
04Are answer engines fetching my documentation?ai answer fetches
05Are training crawlers collecting my content?training crawls

the invisible audience

Your website is already being read by AI.

Most AI crawlers never run browser JavaScript, so script-based analytics never records them. The reading happens every day. It just does not show up.

Traditional analytics

–

Half the audience

HumansSessions, referrers, conversions. All visible.
AI systemsMissing. No script runs, so no event fires.

Armature AI Traffic

✓

The full audience

HumansUnchanged. Keep the analytics you already have.
AI systemsVisible. Every crawl attributed to a system, a bot, a page, and a purpose.

See the AI traffic your analytics cannot.

One prompt to install. Then watch which AI systems read you, page by page.

Start tracking AI traffic → completely free

or talk to Armature