How FaultGuard Diagnoses Equipment Problems in Minutes
Four steps take your technicians from the first symptom observation to a confirmed root cause, a completed repair, and a compliance-ready report — guided by AI every step of the way.
Step 1 — Report What You See, Hear, or Measure
When a technician arrives at the job site, the first thing they do is open FaultGuard, select the piece of equipment from the asset registry, and begin logging what they observe. Symptoms can be entered by choosing from a structured checklist of common indicators — excessive vibration, abnormal noise, high temperature, unusual odor — or by typing a plain-language description that the AI parses into diagnostic inputs.
FaultGuard supports quantitative measurements as well. If a technician takes a bearing temperature reading with an IR thermometer, records a voltage at a test point, or captures a vibration spectrum, those values can be entered directly into the session. The diagnostic engine treats numerical readings as high-confidence evidence, weighting them more heavily than subjective observations when computing probable causes.
Because FaultGuard is offline-first, this entire process works without an internet connection. Equipment models, symptom libraries, and the full Bayesian network are cached on the device. Whether your technician is in a basement mechanical room or a remote pump station with no cell service, symptom reporting works identically.
Photo and audio capture is built in, too. Snap a photo of a corroded fitting or record the sound of a failing bearing, and the evidence is attached to the diagnostic session for later review and reporting.
Step 2 — AI Ranks the Most Probable Causes
As soon as the first symptom is logged, FaultGuard's Bayesian inference engine begins computing. It cross-references the reported observations against the equipment's known failure modes — a model built from manufacturer data, field history, and your organization's own diagnostic experience. Within seconds, the technician sees a ranked list of probable causes, each with a percentage confidence that updates in real time as more evidence is added.
Unlike black-box AI systems that output a single answer, FaultGuard shows its reasoning transparently. Every probability is traceable back to the evidence that produced it. If bearing failure is ranked at 82%, the technician can tap that entry and see exactly which symptoms contributed and by how much. This transparency builds trust, allows experienced technicians to sanity-check the AI, and creates a defensible audit trail for regulated industries.
The probability model is equipment-specific. A centrifugal pump at Station 14 has its own failure history, maintenance record, and operating context. FaultGuard uses per-asset weights with time decay, so a bearing that was replaced two months ago is treated differently from one that has been running for three years. Fleet-wide patterns are incorporated too — if the same model pump is failing across multiple sites, the system raises that signal automatically.
The result is a living, breathing diagnostic model that gets smarter with every session, not a static decision tree that was out of date the day it was printed.
Step 3 — Guided Discovery for Maximum Information Gain
Once the initial probability ranking is established, FaultGuard does not simply hand the technician a list and walk away. Instead, it recommends the single next check that will provide the greatest information gain — the one observation or measurement most likely to separate the remaining hypotheses. This is the difference between systematic diagnosis and trial-and-error guesswork.
The guided discovery engine uses entropy reduction calculations to determine the optimal sequence of tests. If bearing failure and shaft misalignment are the top two candidates, the system might recommend measuring vibration amplitude at the coupling — a result that would strongly confirm one hypothesis while ruling out the other. Each answer the technician provides updates the probabilities in real time, and a new recommendation appears instantly.
Safety is embedded in every recommendation. Before suggesting a check, FaultGuard evaluates the associated hazards and displays danger tags that include required PPE, lockout/tagout requirements, voltage warnings, and skill-level ratings. A junior technician is never directed toward a test that exceeds their qualifications without an explicit safety callout.
This guided loop continues until the system converges on a root cause with high confidence, or until the technician has enough information to make a judgment call. On average, FaultGuard reaches a definitive diagnosis in three to five guided steps — dramatically faster than unstructured troubleshooting, especially for less experienced team members.
Step 4 — Fix with Confidence and Auto-Generate Reports
With a confirmed root cause in hand, FaultGuard transitions from diagnostics to repair guidance. The system pulls step-by-step repair procedures directly from manufacturer manuals, standard operating procedures, and your organization's own documented best practices using Retrieval-Augmented Generation (RAG). Every instruction is cited back to its source document, page, and section — so the technician can verify and the auditor can trace.
Repair guidance includes danger tags at each step where hazards exist: voltage levels, confined space requirements, pressurized systems, chemical exposure risks. The system also flags parts that may be needed, links to inventory if your CMMS integration is configured, and estimates repair time based on historical data from similar sessions across your fleet.
When the repair is complete, the technician marks the session as resolved. FaultGuard automatically compiles the entire diagnostic journey — initial symptoms, probability rankings at each stage, guided checks performed, root cause confirmed, repair steps taken, parts used, time on site — into a standardized PDF report. This report is formatted for regulatory compliance and can be emailed to supervisors, archived in your document management system, or synced to a connected CMMS.
No clipboard. No handwritten notes. No typing up reports back at the office. The documentation that used to take 30 minutes of post-job administrative work is generated instantly, with more detail and consistency than manual entries ever provided.
The Technology Behind FaultGuard
Three core technologies work together to deliver fast, accurate, and continuously improving equipment diagnostics — without requiring cloud connectivity.
Bayesian Inference — Transparent, Editable, Equipment-Specific
RAG — Answers Grounded in Your Manuals
Fleet Learning — Every Diagnosis Makes the System Smarter
Try It Yourself
See the four-step diagnostic process in action with a live walkthrough, or join the waitlist to get early access and help shape the product before general availability.