With over 37,000 GitHub stars and used by teams at Y Combinator, Airbus, and Phantom, PostHog replaces an entire stack of paid analytics tools — Mixpanel, Amplitude, Heap, LaunchDarkly, Hotjar, and Google Analytics — with a single open-source platform where every tool shares a common event layer and user context. Product analytics captures events automatically or via manual instrumentation with HogQL (SQL) access for custom queries, while web analytics provides GA-like dashboards for traffic, conversions, and Core Web Vitals. Session replay records user interactions with DOM snapshots and network waterfall analysis, linking directly to errors and feature flag exposures. Feature flags safely roll out changes to specific cohorts with multivariate support and instant rollback, while experiments run A/B tests with automatic Bayesian significance calculations and revenue attribution. Error tracking captures stack traces linked to session replays and user properties for immediate reproduction context. AI observability monitors LLM generations, traces, token usage, latency, and costs across model versions. The managed data warehouse syncs 120+ external sources including Stripe, Postgres, Salesforce, and HubSpot alongside product events, queryable through a unified SQL editor. An MCP server enables AI agents in Cursor, Claude Code, or VS Code to query analytics and execute SQL directly. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.