beehiivYou built the funnel in PostHog weeks ago. It’s a good funnel.
You look at it when something feels off. Which means you look at it after the drop, not during. The insight was right there, a tab away, for the whole week it mattered.
What the board shows
Your saved insights, run again while the board is open. A trend becomes a chart, like weekly active users or signups per day. A funnel becomes its steps, with how many people made each one. Retention, paths and lifecycle work too, through your agent.
The numbers are PostHog’s own. The board runs the query your saved insight already has, so the funnel on your board and the funnel in PostHog tell the same story.
You ask in plain words. “Put my key PostHog insights on the board” is enough. Your agent looks through your saved insights, including the ones your project views most, and gives each a widget.
Add a sticky note beside the funnel with the one thing you’re fixing this week. It stays in sight until you fix it.
Where the numbers come from
PostHog runs an MCP server, the same one AI apps use to work with your PostHog data. The board signs in to it on PostHog’s own page, and PostHog sends you to your US or EU data.
That server can also create insights and change feature flags. The board never runs those tools. It only runs the reads, while your board is open. If a read fails, the widget keeps the last good number and says what went wrong and when.
Product numbers next to revenue
PostHog tells you what people do in your product. It doesn’t sit next to what they pay. Put weekly actives beside your Stripe MRR and your GA4 traffic, and the whole path from visit to use to payment is on one board.
If agents are doing the building, the AI agent dashboard puts their work beside these numbers too.