// how pitdelta decides
The Pit Board
A pit board says one thing, says it plainly, and only when it matters. PitDelta is built the same way. Five principles decide what a debrief may tell you — and what it must not.
P1
The analysis is computed, not generated.
Encoded telemetry rules select your practice job. A language model never reads your laps, picks a zone, or chooses a number.
P2
One job per stint.
A debrief you can hold in your head at the braking point beats ten graphs you read once.
P3
Not every difference becomes advice.
We look for what repeats across clean laps before we tell you what to change.
P4
No evidence, no assignment.
When the data is weak, PitDelta says so — instead of inventing coaching.
P5
We remember the problem, not just the lap time.
Come back to the same corner and PitDelta tells you whether it moved.
Computed from telemetry. Gated by evidence. No evidence, no assignment.used verbatim on every surface
Computed from 14 clean laps · Engine 7 · build f8369b2printed on every report
One job for the next stint.
Upload a session and see which of these principles fires first on your own laps.
Analyse a session