unknown

Traditional client-side analytics tools miss AI agent traffic to machine-readable docs like `.md` files and `llms.txt` because there's no browser or JavaScript execution involved. This guide walks through building a server-side analytics pipeline using Vercel Log Drains, Hookdeck, and PostHog to capture every such request. Vercel streams server access logs as JSON webhooks; Hookdeck filters batches for `.md`/`llms.txt` requests, transforms payloads into PostHog's batch event format, and handles delivery with retries; PostHog provides dashboards for breakdowns by path, user agent, and region. The transformation uses a djb2 hash of IP, user agent, and JA4 digest for anonymous visitor identity, parses bot names from user agent strings, and emits a custom `doc_view` event to keep server-side traffic separate from client-side pageviews. Limitations include no backfill support, approximate identity resolution, and Vercel Log Drains being a paid feature.