Use Case: Finding the Right PSI for a Pump in the Area 1 Hydrocracker Unit
Let’s say a reliability engineer needs to prepare for a planned maintenance activity on Pump P-1201, one of several pumps in the Area 1 Hydrocracker Unit.
To do the job safely and correctly, the engineer needs:
- The latest maintenance procedure
- The OEM datasheet
- Any MOCs or inspection reports related to this specific pump
- The P&ID showing the pump, associated piping, valves, instrumentation, and upstream/downstream connectivity
- PHAs (e.g., HAZOP or What-If analyses) discussing risks, failure modes, and safeguards tied to that pump or system
These documents provide essential context for:
- Confirming isolation points and pressure boundaries
- Understanding interlocks and control logic
- Identifying any previous risk scenarios or mitigation actions associated with the equipment
- Verifying design assumptions when planning MOC or maintenance
Step 1: Structure the Foundation in SharePoint
Before any AI magic happens, the PSI library needs to be:
✅ Populated with current documents
✅ Tagged with metadata: site, area, unit, system, equipment tag, document type, revision, and effective date
✅ Mapped to a facility asset structure (mirroring the CMMS, like SAP or Maximo)
This creates a structured, machine-readable environment.
Step 2: Train the AI Agent with Domain Context
The AI Agent is then trained to understand:
- Natural language queries and synonyms (e.g., “pump maintenance” = “standard work instruction”)
- The plant’s unique asset taxonomy
- Document types and their functional relevance (e.g., datasheets vs. inspection reports)
Training involves ingesting:
- Sample PSI documents and metadata schemas
- Tag registries or functional location codes
- A glossary of site-specific engineering and safety terms
Bonus tip: Pair this with prior PHA/MOC records to teach the agent how certain risks or equipment types are typically discussed.
Step 3: Deploy the AI Agent Inside SharePoint
The AI Agent can be deployed in several ways:
- As a search assistant embedded in the SharePoint interface
- Through Microsoft Teams integration (e.g., via Power Virtual Agents)
- As a standalone web portal that queries SharePoint APIs behind the scenes
Users can now ask questions like:
“What’s the latest inspection summary for P-1201?”
“Are there any active MOCs involving pumps in Area 1?”
“Show me all procedures tied to hydrocracker pumps from the past 6 months.”
The agent parses the request, navigates the facility hierarchy, filters documents by metadata, and delivers ranked, relevant results—along with context snippets and related files.
How the AI Agent Connects the Dots
Because the AI Agent understands metadata and asset relationships, it can surface related PSI across multiple document types. For example:
Query: “What’s the latest PSI for P-1201 in Area 1?”
The AI Agent may return:
- Maintenance Procedure (Standard Work Instruction)
- OEM Datasheet
- P&ID Sheet U1-451A with tag P-1201 highlighted
- PHA excerpt: HAZOP node for “Hydrocracker Feed Pump P-1201”
- MOC Package MOC-2023-0095, which includes modifications to pump controls and bypass piping
By aggregating all this information, the agent provides a complete picture, so decisions are made based on engineering, operational, and risk-based perspectives.
Step 4: Validate and Learn from Feedback
To refine results, the AI Agent tracks:
- Search terms vs. actual documents accessed
- User feedback on result quality (thumbs up/down, click-through)
- Patterns in how different roles search (e.g., operators vs. engineers)
This learning loop improves accuracy and builds trust in the system.
Business Value Delivered
By combining SharePoint’s robust document management with an AI Agent’s contextual intelligence, you achieve:
✅ Faster access to the right PSI
✅ Reduced downtime during turnarounds or maintenance
✅ Improved safety through fewer human errors
✅ Higher productivity for technical teams
Final Thought
When process safety is on the line—especially in high-hazard units like hydrocrackers—context is everything. Procedures, datasheets, drawings, risk assessments, and change records are all parts of the same story.The FACILEX® PSM suite utilizes SharePoint for all aspects of electronic document management. Gateway can assist with an AI-PSM project for integrating an AI Agent.
