AIWebSignalsobserve the machine web
First-party AI audience intelligence

Follow AI systems through your site.

Turn your own HTTP request logs into evidence-bounded paths: which AI systems arrived, where they entered, what resources they requested next, where they failed, and which sequences repeat. These are observed requests and probable journeys—not access to an agent's hidden prompt or reasoning.

Local-first by designYour raw log file is analyzed in your browser. The journey model drops IP addresses, cookies, query strings, and request bodies. Nothing on this page uploads the selected log file to AIWebSignals.
Journey intelligence

What your AI visitors did

0recognized AI requests
0recognized agents
0probable journeys
0journeys with errors

What this means

  • Import a log or run the sample to produce findings.

Journey map

Your observed agent paths will appear here.
Supported traffic sources

Analyze traffic from the tools you already use

Each provider emits different fields, but AIWebSignals normalizes the same evidence: timestamp, method, host, path, response status, agent identity evidence, and optional request/session identifiers. Every listed source remains locally import-ready. Connected-ingestion maturity is drawn from the canonical connector contract; it does not prove a customer has configured that provider or that live customer traffic has been observed.

Cloudflare Workers

Cloudflare HTTP request JSON or NDJSON.

Workers Trace Logpush -> isolated HTTPS collector

Live validated connector
Vercel

Vercel request logs or drain exports in JSON or NDJSON.

Vercel Custom Drain -> HTTPS collector

Contract validated
Fastly

Fastly real-time logging JSON or NDJSON.

Fastly Real-Time Log Streaming -> HTTPS collector

Contract validated
Akamai DataStream 2

Akamai DataStream 2 JSON or NDJSON.

Akamai DataStream 2 Custom HTTPS -> collector

Contract validated
AWS CloudFront

CloudFront standard access logs with #Fields headers.

CloudFront logs -> Amazon Data Firehose -> HTTPS collector

Contract validated
NGINX / VPS

Default combined access logs.

Local observer/shipper -> HTTPS collector

Local E2E validated
Apache / VPS

Common or combined access logs.

Local observer/shipper -> HTTPS collector

Local E2E validated
WordPress

AIWebSignals WordPress observer request records.

WordPress request observer -> HTTPS collector

Local E2E validated
Other / custom infrastructure

JSON, NDJSON, CSV, or combined-style HTTP logs.

Trusted observer/shipper -> HTTPS collector

Contract validated
Identity confidence

Know what we know—and what we do not.

VerifiedA recognized identity with provider-supplied verification evidence accepted under source-specific rules.
ClaimedA recognized AI identity declared by a field or user-agent string without independent verification.
Probable / unknown automationGeneric automation signals can establish automation without establishing a recognized AI provider. Conflicting identity signals remain unknown.
Inferred sequenceRequests grouped by identity and time without a durable session identifier. Useful for path analysis, not proof of one hidden task.
Use your findings

Turn observed AI traffic into decisions.

Use AI Impact to understand demand and friction, Agent Policy to decide which machine traffic to allow or block, and AI Revenue to price licensed machine access where it makes sense.

Open AI Impact → · Set Agent Policy → · Configure AI Revenue →

Connect first-party trafficOpen connected AI ImpactExplore public Activity Radar

WHY THIS LAYER EXISTS

A human pageview and a machine request are not the same event.

The difference runs from data and protocols through identity, permission, intent, and value. The foundational explainer shows where AIWebSignals fits in that stack.

Understand the human web and the machine web →