How we build things on the internet has changed a lot.
1995 - 2020
The prehistoric days of software development
Analytics, A/B testing, error tracking, and other dev tools required manual implementation using dozens of vendors. (Entire companies were built just around routing data various places!)
2020 - 2024
Multi-product SaaS companies
We started seeing consolidation in B2B SaaS. It became more common to have multiple tools in the same UI.
2025 - current
Just write a prompt
AI now makes it possible to both analyze data and build new features with tooling in place.
Writing code has become easier, but building with AI still has two major flaws:
1. AI-built infrastructure works until it doesn't
It's easy to vibe code a lightweight analytics stack or feature flag system. But it won't scale with any real volume – and querying it gets expensive fast. Your tokens are better spent on your product, not on reinventing PostHog.
2. Context is key
Customer data still lives across various point solutions (database, CRM, support tool, analytics stack). And if you're asking AI to analyze data or write code – its output can only be as good as the context it has.
PostHog solves this in a few ways:
1. Unified data stack
Your data might originate elsewhere, but everything can be pushed into PostHog where it can be transformed, queried, and even exported.
2. MCP
PostHog's dozens of tools are available to your LLM. You no longer need to learn a UI to run analysis or perform tasks like creating an experiment, survey, or feature flag.
3. PostHog Code
Our AI code editor automatically analyzes signals from customer data, proposes improvements, and writes pull requests – automatically.
How we run analysis and build software has changed, but what hasn't changed is the need for good data, good tooling, and a seamless way for them to operate together in harmony.
Try it – free
Use the PostHog Wizard to install PostHog automatically. Each product has a generous free tier – no credit card required.