Introduction
AI-enhanced fault-finding tool is an AI assistant for facility management (FM) technicians that uses a camera, Vision/OCR, and ChatGPT/Claude/Gemini intelligence to identify assets, detect issues, and guide fault resolution on-site.
Proof of Concept Goals and Coverage
Original Goal | How the POC Covers It |
|---|---|
| Technician uses a mobile camera on-site | The Expo mobile app lets the user pick or capture a photo from their phone. |
| Recognize the asset/component in the photo | Mock vision service in the orchestrator simulates object detection + OCR (swap for real Vision API later). |
| Send image/context to ChatGPT backend | The orchestrator builds the Object Analysis prompt from our prompt library and sends it to ChatGPT. |
| ChatGPT performs the "review" | ChatGPT identifies the object, adds function, safety, PM tasks, common faults, and spare parts info in structured JSON. |
| Return actionable response to technician | The mobile app displays all the AI’s findings, including when confidence is low and further images are needed. |
| Fault finding guidance | The same structure supports plugging in the Troubleshooting Guided Flow prompt for step-by-step fault diagnosis. |
| Backend integration | CMMS mock endpoint in the orchestrator shows how logs could be created automatically in your work order system. |
The minimal loop:
capture image → label + OCR → ChatGPT analysis → JSON → display to technician → (optional) send to CMMS.