ML-DiagAssist
Coming Soon
An AI-powered automotive diagnostic platform that pairs a secure OBD-II edge device with cloud-based reasoning and a technician companion app. Think of it as AIOps for vehicles—the same observability and control plane approach used in enterprise IT, applied to automotive diagnostics.
Three-Layer Architecture
Edge Diagnostic Device
A deterministic ESP32-S3 appliance with CAN bus interface that connects to the vehicle's OBD-II port. Handles ISO-TP framing, OBD mode requests, rate limiting, and safety gating—including a physical ARM button required before clearing DTCs. The device is a trusted executor with safety guarantees, not an autonomous agent.
Cloud Reasoning Engine
An MCP server and diagnostic reasoning service in the cloud that normalizes vehicle state, interprets DTCs with AI, produces prioritized test plans, and validates repairs. Enables rapid model improvement without firmware updates and aggregated learning across sessions.
Technician Companion App
A tablet-friendly web app that presents vehicle summaries, AI-explained DTC analysis with severity ratings, guided test workflows, and before/after repair validation. The AI feels like a smart assistant—the technician always stays in control.
Key Capabilities
AI-Powered DTC Analysis
Goes beyond code lookups. The reasoning engine correlates DTCs with live signals, readiness monitors, and known failure patterns to produce natural language explanations and ranked repair suggestions.
Guided Diagnostics
AI suggests tests and the technician approves or denies each step. Commands are routed through the device's policy engine, which enforces rate limits and arming requirements before any write operations reach the CAN bus.
Repair Validation
After a repair, the system compares before and after vehicle state—signals, readiness monitors, and DTC status—to confirm the fix was effective before the vehicle leaves the bay.
Session Evidence Trail
Every session records a complete timeline: signals observed, DTC changes, actions executed, and AI rationale. Invaluable for quality assurance, training, and dispute resolution.
Safe by Design
The edge device enforces deterministic safety policies in firmware. Clearing DTCs requires a physical button press. All CAN bus requests are rate-limited and scheduled. The cloud layer never bypasses the device's policy engine.
Fleet-Ready Architecture
Multi-tenant cloud design supports individual shops and large service organizations. Anonymized, aggregated diagnostic data across sessions enables continuous improvement of AI reasoning over time.
Interested?
ML-DiagAssist is in active development. Contact us to learn more or to be notified when early access becomes available.
Notify Me