AI NativeEngineering Lead

A forecast you can defend, because it shows its inputs

A date range and a confidence signal that cite the sprints they were grounded in

A forecast you can defend, because it shows its inputs

"When does it land?" is the question an engineering lead has to answer in front of people who will remember the answer. A number with no provenance is worthless in that room, which is why PlanMagnet treats a forecast as something that has to cite its sources rather than something that has to sound confident.

PlanMagnet can already do the honest arithmetic. Velocity is computed across the trailing completed sprints of a project, and a next-sprint capacity number is offered from that trailing average. It is labelled for what it is — an average, not a prediction — and it works today because the delivery history and the plan are the same data in the same workspace, with no reconciliation step in between.

The designed upgrade turns that into a real forecast: a delivery window for a project, sprint, roadmap item, or release, with a confidence signal attached, and — the part that matters — a set of input references naming the exact sprint IDs and velocity values the estimate was grounded in. In the reviewing UI those references are intended to be clickable, so a lead can open the sprints behind the date and check the maths before repeating it upward.

It is also designed to fail honestly. A workspace with no completed sprint history gets an explicit "insufficient history" signal, not a plausible-looking date derived from nothing. This is the difference between a forecast and a guess with better typography, and it is the reason forecasting is being built after the grounding data rather than before it.

This is why PlanMagnet can forecast at all: most trackers bolt AI onto a tool whose roadmap lives in another system, so a forecast has nothing local to be grounded in. Here the throughput and the plan share one spine. The deterministic velocity and trailing-average numbers exist in the service today; the grounded, citation-carrying forecast decision is committed roadmap and is described here as design, not as a shipped capability. This story is narrative-only today; the interactive app walkthrough is not yet available.

Ready to make this your story?