F1 Virtual Engineer

F1 Virtual Engineer

Telemetry · Strategy · Explainable AI

Launch
Race engineer · in your browser

Ask any lap.
Get the call.

Compare drivers, replay strategy, read every pit recommendation backed by FastF1 telemetry and a regulation-aware corpus. No spreadsheets.

Open Mission Control on Suzuka 2024 →See a sample query →

4 agents

Multi-agent

pit · tyre · weather · pace

Cited

Every call

linked to FastF1 + regs

Live

Any race

regenerates from lap data

Lap delta

Suzuka 2024 // Race // VER vs HAM

Cited · 3 sources
T1 → finishVER ahead by 0.31s

Sector 1

+0.184

Sector 2

−0.092

Sector 3

+0.221

Lap delta

+0.31s

TYRE MEDIUM
PACE DELTA −0.31
BATTERY DEPLOY PUSH
TYRE MEDIUM
PACE DELTA −0.31
BATTERY DEPLOY PUSH

Featured this week

Where did Suzuka 2024 actually get decided?

Verstappen led every sector by lap 5. The lap-by-lap delta shows when the gap genuinely opened, not just that it did.

Replay the race →

Core capabilities

Telemetry → Decision. No guesswork.

Speed (km/h)
VERHAM

Top

326 km/h

DRS

ENABLED

Δt

+0.31

Telemetry · VER vs HAM
Telemetry + Lap Delta01

Compare two drivers, lap-by-lap, distance-by-distance.

Speed, gear, RPM and per-distance Δt — VER vs HAM at Suzuka, no manual chart parsing.

Pit window — side by sideCited
VERLOW

Window

L18 → L22

MED → HARD

HAMMED

Window

L20 → L24

MED → HARD

Strategy · what-if
Strategy Compare02

What-if pit calls, side by side.

Pit-window recommendations with target-driver gap context, tyre features, and rationale you can read.

VER · Suzuka
20242023
Best lap delta−0.42s · 2024 faster
Cross-year · Δ fastest lap
Cross-Year + Citation03

Same driver, two seasons, one delta line.

Compare a driver's fastest lap year-over-year with citations from a regulation-aware corpus.

◐ MIXED · 23°C trackMED · 14 lapsDEG · HIGH

Compound history

Header context
Weather + Tyre Context04

Conditions in the header, compound history in the card.

Per-session weather pill (DRY / MIXED / WET) with tyre framing — the context strategy actually depends on.

For the curious

Built on real lap data.

Every recommendation regenerates from FastF1 lap data and a regulation-aware corpus. No hand-tuned numbers, no "trust me" metrics — every call traces back to a source.

  1. 40%

    Baseline

    Pit-window heuristic shipped against a 5-fixture seed set. Useful enough to demo, but the pit-lap calls drifted on long stints.

    01

  2. 65%

    Tier rework

    Re-bucketed the tyre-deg tiers and pinned the window to lap-count, not constants. CI snapshot expanded to 20 fixtures.

    02

  3. 92%

    Live gate

    25-fixture snapshot now gates merges. Failures regenerate from FastF1 lap data; rationale is grounded in a regulation-aware corpus.

    03

Pit wall awaits

Step into the pit wall.

One demo session loaded. Pick any race after that — every call is cited, every chart redraws from lap data.

Launch Mission Control →