A new Insights tab inside the Tasks page shows you which services are running over or under estimate. It's powered by the actual hours your crew has been clocking on jobs (since the clock-in feature shipped).
Three sections: services running over (red), running under (green), and "not enough data yet" (sparse). Click any service to see the breakdown — which tasks specifically are off, and which projects pulled the median up or down.
Below the calibration card on each service page, a new Used on projects card lists the last 5 signed projects that used the service, so you can jump straight to the underlying job whenever you want context.
This is the read-only first step in the calibration loop. AI-suggested fixes are coming next — once a service accumulates 5+ completed jobs with consistent drift, BidCraft will propose a specific change to your service template that you can accept or reject with one tap.