Not a plan generator. Not another dashboard.
Most running software falls on one of two sides. Plan generators hand you next week’s workouts without ever understanding what happened in today’s: the plan adapts, but nobody explains. Chart dashboards do the opposite: forty numbers about this morning’s run and not one sentence about what they mean. Between the two sits the thing a coach actually does after a run: look at what happened, work out why, and say what to do next. That’s a debrief, and it’s the whole product.
What it actually reads
A debrief is only as good as its evidence. debrief takes the FIT file straight off your Garmin or COROS watch and decodes it losslessly, then rebuilds the run as a second-by-second record with every doubtful stretch flagged. A GPS dropout is marked as a dropout, not smoothed into fiction. Onto that record it layers the day’s weather, dew point included, and the terrain under you from a satellite elevation model. Treadmill, trail and track all count. The pipeline runs against real COROS and Garmin recordings on every change.
It speaks runner, not dashboard
Every pace in a debrief reads m:ss/km. Metrics never lead a sentence; meaning does, the way a coach would text you: “your training balance is −13.9 (TSB)”, never “TSB sits at −13.9 with ATL above CTL”. No sentence carries more than two metrics, on the theory that a third means the sentence should have been two.
And when a run was simply the run you meant to do, the debrief says so in a few sentences and stops. Silence beats filler. A coach who finds something profound in every Tuesday jog is a coach you stop reading.
Honest about what it doesn’t know
An AI coach earns trust by admitting the limits of its evidence. The load ratio marketed elsewhere as an injury predictor is labelled what it is, contested science, and never carries a conclusion alone. The heat model that learns your personal response to humidity says nothing until twenty-five sessions across a real spread of dew points back it up: no number is better than a fake number. And every comparison is against your own recent history, never a population average; your easy pace measured against a stranger’s would be noise.
What a debrief covers
Each run’s stress is scored from grade-adjusted pace, with heart-rate fallbacks when GPS can’t be trusted, and rolled into your training load: long-term base, short-term load, training balance. Long runs get read for durability through aerobic decoupling. And every judgment is weighed against the conditions, because dew point changes what a split means. What happened, why, and what to do next, after every run.
Get your first debrief
debrief is in early access. Leave your email and your first uploaded run gets the full treatment.