What do I configure on the Fastly side?

An HTTPS logging endpoint on the service you want to track, pointed at your LLM Pulse ingest URL. It ships with your next service version activation, with no VCL changes required.

What log format should I use?

Newline-delimited JSON (NDJSON), including timestamp, user agent and URL fields. That is all LLM Pulse needs to identify and attribute AI bots.

Will this affect caching or delivery performance?

No. Fastly's logging pipeline runs out-of-band from request handling, so delivery latency and cache behavior are unaffected.

Which plans include the Fastly integration?

Agent Analytics is included on Scale and Enterprise plans, with per-URL drilldown on Enterprise.

NEW ChatGPT Ads is live : track paid ad placements in ChatGPT Learn more

Agent Analytics Integration

Stream AI bot traffic from Fastly in real time

Use Fastly Real-Time Logs to stream request data to LLM Pulse and identify every AI agent crossing your edge, from GPTBot to PerplexityBot, as it happens.

Start Free Trial

14-day free trial

Cancel anytime

Features

Real-time edge logs, real AI agent visibility

Fastly's logging pipeline is one of the fastest in the industry. Point it at LLM Pulse and your AI bot picture is always current.

Real-time HTTPS log streaming

Create an HTTPS logging endpoint in Fastly, paste your ingest URL, and logs arrive as newline-delimited JSON seconds after each request.

Every AI agent identified

Requests are matched against our catalog of AI bots, including GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot, Google-Extended, Bytespider and Amazonbot, grouped by company and category.

Works per service

Attach the logging endpoint to whichever Fastly services you want to monitor and keep AI bot tracking scoped to the right properties.

Tokenized endpoint with rotation

Your ingest URL carries a unique secret. Rotate it whenever you need, with a 48-hour grace period for the old URL so no logs are dropped.

How the Fastly integration works

One logging endpoint, configured once.

Generate your ingest endpoint

From Agent Analytics, choose Fastly Real-Time Logs and generate your unique ingest endpoint URL.

Add an HTTPS logging endpoint in Fastly

In your Fastly service, go to Logging, create an HTTPS endpoint, paste the URL and choose newline-delimited JSON as the format.

Deploy and observe

Activate the service version and logs start streaming. AI bot totals, time series and company breakdowns appear in your dashboard within minutes.

"LLM Pulse has allowed us to better organize our work and prioritize key actions. It gives us much greater clarity when implementing new ideas and makes reporting easier, with metrics that were previously complex and difficult to obtain."

Sandra Vargas

Euroinnova

Frequently Asked Questions

Common questions about the Fastly integration for Agent Analytics.

: An HTTPS logging endpoint on the service you want to track, pointed at your LLM Pulse ingest URL. It ships with your next service version activation, with no VCL changes required.
: Newline-delimited JSON (NDJSON), including timestamp, user agent and URL fields. That is all LLM Pulse needs to identify and attribute AI bots.
: No. Fastly's logging pipeline runs out-of-band from request handling, so delivery latency and cache behavior are unaffected.
: Agent Analytics is included on Scale and Enterprise plans, with per-URL drilldown on Enterprise.

See AI agents crossing your Fastly edge

Stream real-time logs to LLM Pulse and track every AI bot. Live on Scale and Enterprise plans.