HPFP · 142/200 bar UNDER LOAD
P0300 · MISFIRES FOLLOW DEMAND
LPFP IN-TANK · ✓ RULED OUT
> P0300 ACTIVE — TRACING_

Freeze frame in.
Test plan out.
90 seconds.

An active diagnostic engine built for the bay. It calculates the one test worth running, reads the waveform you capture, and re-ranks every suspect before you've set the scope down.

Scanning is free forever, never metered — works with the $15 ELM327 in your drawer, in Chrome. No card, no proprietary hardware, no unlock fees.
90seconds to a ranked test plan
15dollar adapter, not $4,000
0dollars to scan, on every plan
SCROLL TO SCAN
01 · Scan the bus

2022 BMW M240i. Random misfires under load. Two sets of coils already.

P0300, intermittent, only above 4,500 RPM. Two shops threw plugs and coils at it — twice. Watch Lane Pilot roll-call the bus and chase the misfire past the ignition system everyone already replaced.

  • DME⚠ P0300
  • EGS
  • DSC
  • FEM
  • ACSM
  • KOMBI
  • IHKA
  • ZGW🔒 GUARDED

THE DIAGNOSTIC WORKFLOW

01 · SCAN

Roll-call the bus — and flag what it can’t reach

Connect your $15 OBD-II dongle. Lane Pilot discovers every module that answers, captures freeze-frame data and pulls live PIDs in seconds — then names the ones a secure gateway kept it out of, instead of calling the car clean.

Web Bluetooth · Any ELM327
02 · DIAGNOSE

Specialist agents reason in parallel

Librarian, NHTSA and Shop Data retrieve at once; the Mechanic synthesises and the Foreman runs the safety pass. Result: a cited, ranked test plan in 30–90 seconds.

Multi-Agent · All Makes · 30–90s
03 · VERIFY

Prove the fix before the customer drives away.

Re-scan after the repair, then arm I/M Monitor Watch — it polls the readiness monitors every 5 seconds and beeps when they all set. Evidence on the RO.

I/M Monitor Watch · Mode 01 PID 01
02 · 90 seconds to a test plan

Not a code definition. A step-by-step test plan.

Built from this car's live data, real-world failure patterns, and recall records — every suggested repair prompting the tech for video evidence on the RO.

DIAGNOSTIC ENGINE · 2022 BMW M240i B58 · T+84s
P0300 random misfire, only under hard acceleration. Two shops already did plugs and coils — twice. Freeze frame attached.
engine reasoning…
This isn't ignition. Freeze frame: misfires cluster above 4,500 RPM at >80% load. Coil failures don't follow demand — fuel delivery does.Verdict: rail pressure 142 bar vs 200 bar commanded at WOT → 29% below demand [✗ OUT OF RANGE — fleet baseline for this engine: ±4%]1. Log rail pressure on a 3rd-gear pull. Sags with demand = high-pressure side. LPFP delivery already in range — ruled out.2. Command a rail-pressure setpoint via 0x2F and watch the pump respond — confirms HPFP without dropping the tank.3. NHTSA recall check: none active for this VIN.⚠ Upload video evidence of the coil swap to attach to the RO.
POST-REPAIR VERIFY: LIVE RAIL PRESSURE vs COMMANDED — HPFP TRACKING✓ VERIFIED
> 0x2F RAIL-PRESSURE SETPOINT → NRC 0x33 SECURITY ACCESS DENIED
> SGW GUIDANCE CARD → NASTF / BMW AOS TOKEN PATH .......... SHOWN (fail-closed, no key guessed)
> LIVE RAIL PRESSURE AFTER REPAIR → WATCHING
WOT DEMAND: +60 barRAIL TRACKS DEMAND — NOMINAL
03 · Fixed means verified

The car said no. Lane Pilot told you exactly what it wants.

New pump in. The gateway rejected the command — so Lane Pilot named the NRC, showed the SGW guidance card for your shop's token path, and streamed live rail pressure so you can watch the new HPFP track. No "should be fixed." The RO carries the evidence — and the two coil jobs it didn't need.

Everything you just watched ran in a browser tab.
On a $15 adapter.

No tablet. No proprietary VCI. No unlock fees.
The other way to do this costs $4,000 and lives in a locked drawer.

Web BluetoothELM327 CompatibleNo Install RequiredOffline-Tolerant

THE PLATFORM

Next-Best-Test Engine

Active Bayesian reasoning. Calculates which physical test eliminates the most possible causes per minute of labor.

Waveform Intelligence

Drop scope captures straight into the chat — .CSV export or a screenshot. The engine reads the signal and re-ranks the suspect list, same turn.

Reasons, never guesses

Deterministic math. Real telemetry. The engine reasons on numbers it cannot invent.

Fail-closed actuation safety

IOControl 0x2F overrides run behind a fail-closed allowlist and a VSP-gated security-access engine, every request audited. The command catalogue is being rebuilt against capture-verified sources.

CAN bus monitor

29-bit capture, live module roll-call, CSV + PEAK .trc export.

Offline-first

Writes locally, syncs when WiFi allows. Dead zones in the bay survive.

Enterprise API

POST /v1/diagnose → cited plan in 30–90s. Any scanner, any platform.

Acoustic NVH

Phone-mic envelope analysis. Anchored to RPM, it splits 0.5× valvetrain from 1.0× bottom-end — and stays quiet when it can't be sure.

Patterns. Not promises.

worked examples · not shop data

PATTERN FIX

2022 BMW M240i B58 · P0300 under load → high-pressure fuel pump

example · platform match
PATTERN FIX

2018 Ford F-150 3.5L EcoBoost · P0016 cold rattle → cam phasers, not the chain

example
RULED OUT

P0300 under load on B58: plugs & coils — wrong diagnosis

anti-pattern

Every diagnosis feeds the loop. Every confirmed fix makes the next one faster.

Per-VIN diagnostic history

Your scan tool forgets. This one doesn't.

A dealer network keeps a scan history on every vehicle it services — one brand, on their side of the counter. An independent bay starts from zero. LanePilot closes that gap: every scan is saved against the VIN — a per-vehicle diagnostic database your shop builds automatically. Yours until you delete it.

VEHICLE HISTORY · SAME VINrepresentative example · not shop data
VISIT 1 · MAR 14 · 87,412 mi
LTFT B1 +2.1% · MAF 4.9 g/s · timing 12°
✓ baseline recorded for this VIN
VISIT 2 · AUG 22 · 91,208 mi · +3,796 mi
LTFT B1 +9.4% — drifted +7.3 vs this vehicle's baseline (threshold ±5%)
CalID changed since last visit — reflashed or tuned
codes cleared 11 mi / 2 warm-ups before this scan
PRIOR VISITS

The Diagnostic Engine reads the last five visits — the complaint, the codes that were stored, and the fix.

CALID WATCH

Flags a calibration change since the last visit — reflashed or tuned.

BASELINE DRIFT

Trims, MAF, timing and O2 graded against that vehicle's own prior sessions.

CLEAR RECENCY

How many miles and warm-up cycles ago the codes were cleared.

The first scan of a vehicle establishes its baseline — the record starts paying off the second time it comes in.

The language model never does the math. Margins, limits, and z-scores are computed in code and handed to the model as ground truth it cannot change — so every verdict is deterministic, auditable, and repeatable.

PLATFORM ARCHITECTURE · VERIFIABLE · NOT A CLAIM
Your worst car next

Bring us your worst intermittent.

Plug in what you already own. First scan in minutes.

A comeback eats hours you can't bill — diag time, the redo, and the customer you lose. One avoided comeback covers months of Lane Pilot.