Per-VIN circuit interpretation, AI fault-tree diagnostics, and pin-out reasoning grounded in your wiring data. Snap a photo of the harness — the OCR layer reads wire colors, connector labels, and pin numbers off the picture so the assessment matches what is actually in front of you. Built for technicians chasing real electrical complaints — no-crank, parasitic draw, lamp-out, CAN-bus glitches, and intermittent shorts.
It is not a chatbot draped over a generic LLM. The Assistant is grounded in the per-VIN circuit context already in TemplateVault — connector IDs, pin assignments, wire colors — and reasons about it the way a journeyman tech would.
Pulls the circuit table for your VIN and lays out the affected circuit before suggesting the next test. Nothing generic, nothing year-agnostic.
Snap a picture of the harness, connector, splice, or burned wire. The OCR layer reads wire colors, connector callouts, and pin numbers off the photo and reconciles them against the circuit table — so the diagnosis is grounded in what is in front of you, not a generic schematic.
Pick a complaint or paste a customer description. The Assistant proposes a working hypothesis, the next test, the expected reading, and what each result implies.
Ask "where do I back-probe pin 7?" or "what is the wire color for the rear washer relay coil?" — answers reference the circuit table for that VIN and any photos you have attached.
Step-by-step current-draw isolation, voltage-drop test plans, and a clear path from "battery low after 3 days" to the actual offender.
Reads the lighting circuit for that VIN, points at the most likely failure mode (bulb, ground, BCM driver, fuse), and lays out the test sequence to confirm.
Walks resistance checks across the bus, identifies likely terminating-resistor or wiring concerns, and prompts when to reach for the scope.
Photograph a chewed, melted, or cut harness and the OCR layer pulls every legible wire color and label out of the picture, then maps them back to the circuit so you can scope the repair before you start cutting.
Five steps from a customer complaint to a confirmed root cause.
Decode happens via NHTSA on first save. The circuit table for that VIN is already linked to the vehicle record.
Choose from common patterns (no-crank, parasitic draw, lamp-out, CAN-bus, intermittent short) or paste the customer description and the Assistant classifies it.
Optional but powerful. Shoot the connector, splice, or burned section in the bay. OCR pulls wire colors, connector labels, and pin numbers off the picture and reconciles them with the circuit table for the VIN — so the assessment matches what you are actually looking at, not a generic schematic.
The Assistant lays out the relevant connectors, pins, and wire colors before suggesting any test. Photo evidence is highlighted alongside the schematic so you always see what it is reasoning over.
Next test, expected reading, and the implication of each result. Step-by-step until the bad component reveals itself.
Confirmed fault, test readings, photos, and parts replaced are pushed into the AI Work Order Narrator so the customer-facing write-up is honest and complete.
We label outputs clearly. The Wiring Assistant is genuinely useful — and we will not dress AI synthesis up as OEM data.
AI inference: Google Cloud Vertex AI by default, with Claude as fallback when Vertex is unavailable or the request benefits from Claude's strengths.
Grounding: the Assistant reasons over the per-VIN circuit table already in TemplateVault. Where I-CAR has published guidance for the procedure, results include direct source links to rts.i-car.com so you can verify.
Labeling: every output carries an AI-Assisted badge or a Source-Referenced badge. You always know which one you are reading. The direct OEM data pipeline is in active development.