Book a demo

05 / 07 — DiagAIAvailable

AI vehicle diagnostics software.

Ask your measurements a question; agents answer from your own data.

Eliminate endless troubleshooting trees. Route vehicle DTCs, freeze frames, and live CAN telemetry through 8 specialized diagnostic agents covering ISO 14229 UDS, OBD-II, J1939, CAN and LIN — deployable fully on-premise, with no data leaving your network.

Book a DiagAI demo See pricing

In pilot with Tier-1 powertrain teams

DiagAIcase_P0302.diag Live
Evidence P0302Cylinder 2 misfire
  • freeze frame · 3 480 rpm
  • run_07.mf4 · 1 ms raster
  • lambda bank 1 lean
MisfireAgent
FuelAgent
DFCAgent
IUPRAgent
DrivabilityAgent
Ranked root cause
Ignition coil, cyl 20.72
Injector cyl 2, low flow0.18
Crank sensor dropout0.06

Next: swap coils 2↔3, re-run misfire counter

UDS · OBD-II · J1939 · on-premise agents · illustrative case
Protocols
ISO 14229 (UDS), OBD-II, SAE J1939, CAN, LIN
Agents
8 specialized diagnostic agents
Local inference
LM Studio / Ollama-compatible (e.g. DeepSeek)
Isolation
Per-OEM RAG index, offline-first

Capabilities

What DiagAI does.

  1. E1

    Multi-Agent Collaborative Reasoning

    Misfire, Fuel, Thermal, Combustion, and Bus Integrity agents cross-examine telemetry to pinpoint subtle intermittent faults.

  2. E2

    Protocol-Native Context

    Deep understanding of UDS Service $19 (ReadDTC), $22 (ReadDataByIdentifier), snapshot records, and extended freeze frames.

  3. E3

    Signal Intelligence & Alias Matching

    Automatically maps proprietary OEM signal names to standardized diagnostic physical channels across legacy DBCs.

  4. E4

    Per-OEM RAG Isolation

    Runs on your local workstation or private data center via LM Studio or Ollama-compatible models. Each organization's data and documentation stays in a separate local index.

8-AGENT SPECIALIZED NEURAL REASONING MATRIX

Domain-expert agents running concurrently over ISO 14229 UDS and SAE J1939.

Multi-Agent Diagnostic Specialist Swarm

Rather than relying on a generic LLM, DiagAI routes vehicle telemetry through 8 domain-specialized agents that collaborate in parallel:

MisfireAgent AGT-01

Analyzes cylinder-individual crankshaft acceleration drops, ionization metrics, and Mode test IDs to distinguish ignition breakdown from injector clogging.

FuelAgent AGT-02

Tracks high-pressure fuel rail transients, STFT/LTFT adaptations, and pump duty cycle to isolate pressure regulator wear and fuel delivery anomalies.

WLTPAgent AGT-03

Evaluates driving cycle telemetry against WLTP and Real Driving Emissions (RDE) boundaries to flag cycle drift and catalyst light-off delays.

DFCAgent AGT-04

Traverses Diagnostic Fault Code dependency trees, debouncing counters, and ECU healing criteria to separate primary faults from secondary cascade codes.

IUPRAgent AGT-05

Audits In-Use Performance Ratios across catalyst, evaporative system, and secondary air monitors for OBD-II regulatory compliance.

KnockAgent AGT-06

Performs acoustic knock sensor window analysis and cylinder ignition retard decay inspection to detect super-knock and pre-ignition risks.

ThermalAgent AGT-07

Monitors coolant circuit temperature gradients, electric water pump flow rates, and charge-air cooler effectiveness across ambient extremes.

EmissionsAgent AGT-08

Inspects upstream/downstream NOx sensor correlation, SCR urea dosing models, and DPF differential pressure soot loading curves.

ISO 14229 (UDS) Protocol Diagnostic Routing

DiagAI interfaces natively with Unified Diagnostic Services over CAN, CAN-FD, and DoIP (ISO 13400):

Service ID (SID) Service Name Standard Function DiagAI Autonomous Operation
0x19 ReadDTCInformation Sub-func 0x02 (ByMask), 0x04 (SnapshotRecord) Extracts DTC freeze frames and environmental freeze-frame data tables
0x22 ReadDataByIdentifier Dynamic DID polling (0x0100 – 0xFFFF) Continuously queries high-speed live sensor arrays without bus saturation
0x2E WriteDataByIdentifier Variant coding and adaptation value writing Applies safe adaptation resets with automated rollback protection
0x31 RoutineControl Actuator actuation routines Orchestrates safe automated component functional test sequences

Workflow

From input to decision.

  1. 01

    Collect Evidence

    Import DTC freeze frames, CAN traces, and driver symptom descriptions into DiagAI.

  2. 02

    Agent Analysis

    Specialized agents evaluate evidence simultaneously, calculating fault probabilities and correlating sensor anomalies.

  3. 03

    Prescribe Action

    Receive an actionable, ranked root-cause diagnosis and a guided step-by-step physical verification procedure.

Why it exists

The veteran diagnostician embedded in software.

DiagAI does not produce generic chatbot summaries. It performs structured, protocol-aware diagnostic reasoning backed by automotive standards, isolating root causes in minutes rather than days.

Each agent operates on MDF/MF4 signal data and returns structured findings — cross-referencing diagnostic trouble codes against live measurement signals for root-cause analysis.

Agent capability matrix

Eight agents, each with a domain.

Specialized agents cross-examine the same evidence from different diagnostic angles.

AgentDiagnostic domainProtocol coverage
MisfireAgentCombustion analysis, cylinder-level anomaly detectionUDS, OBD-II
FuelAgentBSFC / BMEP evaluation, long-term fuel trim, injector healthOBD-II, J1939
WLTPAgentEmissions compliance, regulatory drive-cycle simulationOBD-II, CAN
DFCAgentFault code interpretation, DTCs, OEM-specific DFC logicISO 14229 UDS
IUPRAgentOBD-II monitor readiness, in-use performance ratioOBD-II
GearAgent, DrivabilityAgent, StartWarmupAgentGear hunting, drivability, start / warm-upCAN, LIN

Questions

Asked by engineers, answered plainly.

Which agents handle misfire detection diagnostics, and is there a WLTP data analysis tool?

DiagAI includes 8 specialized agents: MisfireAgent for combustion analysis, FuelAgent for BSFC and BMEP evaluation, WLTPAgent for emissions cycle analysis, DFCAgent for fault code interpretation, IUPRAgent for OBD-II monitor readiness, GearAgent for gear-hunting detection, DrivabilityAgent, and StartWarmupAgent. Each agent operates on MDF/MF4 signal data and returns structured findings.

Can DiagAI run completely on-premise without sending data to the cloud?

Yes. DiagAI supports fully on-premise deployment via LM Studio or Ollama-compatible local LLMs such as DeepSeek. Per-OEM RAG isolation means each organization's data and documentation stays in a separate local index. No data leaves your network unless you explicitly configure a cloud inference endpoint.

Is DiagAI a UDS ISO 14229 diagnostics tool and J1939 data analysis software?

DiagAI supports ISO 14229 (UDS), OBD-II, SAE J1939, CAN, and LIN. It parses diagnostic trouble codes and live signal data across these protocols, then cross-references them with MDF measurement signals for root-cause analysis.

Does DiagAI require an internet connection to function?

No. DiagAI is designed for offline-first operation. All signal parsing, agent inference, and RAG retrieval run locally. An internet connection is only needed if you configure an external cloud LLM endpoint, which is optional.

Next step

Ready to augment your diagnostic lab?

Eight protocol-aware agents, running on your own hardware, reasoning over your own measurement data — with every OEM index kept separate.

Book a DiagAI demo Talk to an engineer

By task

Where DiagAI is used.

Go deeper

Guides and comparisons for DiagAI.

Works with