IFS Cloud + maintenance AI
AI for IFS Cloud Maintenance: Work Orders, PM Plans and Spares Answered in Plain English
Short answer
AI for IFS Cloud maintenance lets planners, reliability engineers and technicians ask plain-English questions across work orders, equipment objects, preventive maintenance plans, spare parts and fault history. It reads IFS data through its APIs with the asker's own permissions, shows the query behind each answer, and can run on your own infrastructure.
- ERP
- IFS Cloud, IFS Applications 10
- Industries
- Aerospace, Defense, Utilities, Heavy Industry
- Written for
- Maintenance Manager
IFS Cloud holds a deep maintenance record: functional and serial objects with their structures, work requests and fault reports, work orders and work tasks, preventive maintenance actions, labor and material reporting, and the spare parts that tie it to inventory and purchasing. Maintenance managers know the data is there. What they lack is time to assemble it. A question like which pumps had more than two corrective jobs in a year, and what parts did those jobs consume, usually means a Lobby element someone built years ago or an export to a spreadsheet.
The result is that the most valuable maintenance questions are asked rarely. Backlog by craft and priority gets reviewed weekly because someone prepared it. Why a specific asset keeps failing gets discussed from memory, because the join between fault reports, work task history and part consumption is tedious. Planners spend mornings searching for the last time a job was done, and technicians describe the same symptom in free text in five different ways.
A private AI layer sits beside IFS Cloud and reads from it. It does not replace IFS work order management, the planning boards or your approval flows. It answers cross-object questions quickly, summarizes long fault histories, drafts work descriptions from past jobs, and flags PM plans that no longer match how the asset actually fails. Anything that changes a record stays behind human approval inside IFS.
This page explains where AI fits in IFS Cloud maintenance, how it connects, what it should not do, and what to ask any vendor, including us. If your need is aircraft maintenance records under EASA or FAA rules, see the aviation MRO page. If you run IFS in the field service context, see the field service page. This one is about plant and asset maintenance.
What usually gets in the way
The problems we hear most from maintenance manager teams running IFS Cloud.
Failure history is spread across objects
Fault reports, work orders, work tasks and reported material each hold part of the story for an asset. Finding repeat failures means joining them, and the free-text descriptions rarely use consistent words.
PM plans drift from reality
Preventive maintenance actions are set up once and revised rarely. Intervals stay fixed while actual failure patterns change, so some tasks are over-performed and others miss the failure mode that keeps happening.
Backlog review is manual
Planners reconcile open work orders against craft capacity, parts availability and shutdown windows by hand. Jobs waiting on parts are found late, when a shutdown is already scheduled.
Spares decisions lack asset context
Min and max levels in inventory are often set from consumption alone. Whether a part is critical to a specific functional object, and how long it takes to get, is known to people, not to the reorder logic.
Knowledge walks out the door
Experienced technicians know that a vibration alarm on a certain gearbox usually means a coupling problem. That knowledge lives in old work order text and in heads, and new staff cannot search it.
Where AI earns its place in IFS Cloud
Each use case names the ERP objects it reads or writes, so your ERP team can judge the integration effort before anyone commits budget.
Repeat failure analysis by asset
A reliability engineer asks which assets had three or more corrective work orders in twelve months, grouped by failure code and cause, and gets a table with the queries shown and links back to the work orders.
Touches: Functional Objects, Serial Objects, Fault Reports, Work Orders, Work Tasks, Failure Codes
Outcome: Reliability reviews that took half a day to prepare reduce to a short check of the output. The failure code quality still decides how useful the grouping is.
Backlog and shutdown readiness
The agent lists open work orders for a planned outage by craft, estimated hours and part availability, and highlights jobs with unreserved or backordered material.
Touches: Work Orders, Work Task Planning, Material Requisitions, Inventory Part in Stock, Purchase Orders
Outcome: Parts gaps are found weeks earlier than at the job kickoff. The check is only as good as material planning on the work orders.
Work description drafting from past jobs
Given a fault report, the assistant finds similar closed work orders, drafts a task description with likely steps, craft and parts, and the planner edits and approves it.
Touches: Fault Reports, Work Orders, Standard Jobs, Reported Material, Reported Time
Outcome: Planners spend less time writing and searching. Output is a draft; standard jobs and safety instructions remain the controlled source.
PM interval review
The agent compares PM actions against corrective history for the same objects and flags plans with no findings over many cycles, or objects failing between PM visits.
Touches: PM Actions, PM Plans, Work Orders, Maintenance Programs, Equipment Objects
Outcome: A ranked list of PM plans worth a second look, with evidence. Interval changes still go through your reliability and approval process.
Critical spares check
Ask which parts used on critical functional objects have no stock and open lead times beyond a threshold, with the objects they support.
Touches: Inventory Parts, Part Catalog, Supplier per Part, Object Spare Parts, Purchase Order Lines
Outcome: Spares risk becomes visible by asset criticality rather than by consumption alone.
Technician history lookup
A technician on a mobile device asks what was done last time this alarm appeared on this machine and gets a short summary with the source work orders cited.
Touches: Work Orders, Work Task Descriptions, Documents attached to objects, Object Structure
Outcome: Faster first-time diagnosis for newer staff. Answers cite the source records so they can be checked.
Cost and downtime reporting
Managers ask for maintenance cost per object or per site split into labor, material and external services, with corrective versus preventive share.
Touches: Work Order Cost, Posting, Reported Time, Reported Material, Maintenance Cost Structure
Outcome: Ad hoc cost questions answered without waiting for a report build, with the financial figures traced to postings.
Reference architecture
The AI layer connects to IFS Cloud through its standard REST projections and, where licensed and appropriate, read replicas or reporting views. Retrieval is built on structured queries plus an index of free-text fault and work descriptions. Models are open-weight and run on your own GPUs, so maintenance and asset data never leaves your network.
- 1
IFS data access
OData-style REST projections for work orders, objects, parts and inventory, authenticated with an IFS service identity or delegated user identity. Optional read-only replica for heavy analysis so production users are not affected.
- 2
Permission and context layer
Questions run under the asker's IFS permissions and site or organization scoping, so a planner at one site does not see another site's records unless IFS would allow it.
- 3
Retrieval and text index
Free-text fault descriptions, work task notes and attached documents are indexed so similar-job search works on meaning, not exact words. Index lives in your environment.
- 4
Model serving
Open-weight models served with vLLM or Ollama on your GPUs or private cloud. Generated queries are shown to the user before and after execution.
- 5
Action and approval layer
Write-back, such as creating a draft work request, is off by default. When enabled it posts a proposed change that a named person approves in IFS, and every action is logged.
Integration notes for your ERP team
- Use IFS Cloud REST projections for reads. Confirm which projections are exposed for work orders, objects and parts in your release and what your hosting arrangement allows.
- Service accounts need IFS permission sets matching the data scope, not a blanket admin identity. Review them with your IFS administrators.
- For heavy trend queries, a read-only replica or reporting copy avoids load on the transactional database. Whether this is available depends on your hosting model.
- Failure codes and object structure quality drive answer quality. Plan a short data audit of failure code usage before relying on repeat failure analysis.
- Documents attached to objects can be indexed for retrieval, but access rules on document classes should be mirrored in the index.
- If IFS Lobby or existing BI reports already answer a question well, keep them. Use AI for the cross-object and free-text questions they handle poorly.
- Mobile or Aurena access to the assistant can be provided through your existing identity provider so technicians sign in once.
Deployment options
On-prem or private cloud next to IFS
Operators of critical infrastructure, defense sites and plants with strict data policies
Models, index and orchestration run on your servers. IFS Cloud connectivity is configured per your IFS hosting model, which should be confirmed early because it determines API reachability.
Sovereign or regional cloud
Organizations that need in-country processing but do not run GPUs
GPU hosting in a region and tenant you control, with private networking to IFS. Suitable when data residency matters more than owning hardware.
Read-only pilot on one site
Teams that want evidence before a wider rollout
A single site, a handful of asset classes and read-only access, running for a few weeks so planners can judge answer quality on real questions.
Compliance and data control
How the architecture supports your obligations. Certification and accountability stay with your organisation; the design keeps the evidence straightforward.
Asset and safety records
AI output is advisory. Safety-critical maintenance decisions and sign-offs stay in IFS under your procedures, and the assistant never closes or releases work.
Access control
Queries inherit IFS permissions. Service identities are scoped to read the objects needed and reviewed like any other integration account.
Data residency
On-prem or in-region serving keeps maintenance and supplier data under your control. Netray does not need a copy of production data.
Audit trail
Each question, generated query and answer is logged so reviewers can reconstruct what was shown to whom.
ISO 55001 style asset management
The architecture can support your asset management evidence by making history and cost queries repeatable. Certification remains your organization's responsibility.
Where Netray fits
How an engagement runs
Phase 1 . 1 to 2 weeks
Discovery
- -Review of IFS hosting model and API access
- -Inventory of maintenance questions that cost the most time
- -Data quality check on failure codes and object structures
- -Security and permissions plan
Phase 2 . 3 to 5 weeks
Read-only pilot
- -Connection to IFS maintenance projections
- -Assistant for backlog, failure history and spares questions
- -Accuracy review with planners on real questions
- -Query log for audit
Phase 3 . 3 to 4 weeks
Text and document layer
- -Index of fault and work task text
- -Similar-job search and draft work descriptions
- -Document retrieval with access rules
- -Evaluation set of known answers
Phase 4 . 2 to 4 weeks
Production and handover
- -Deployment on your infrastructure
- -Monitoring and cost reporting
- -Optional approved write-back for draft work requests
- -Runbook and training for maintenance teams
Questions to ask any vendor, including us
A short list that separates real IFS Cloud AI work from a chatbot demo.
- How does the assistant authenticate to IFS Cloud and which permissions does it hold?
- Can I see the exact query behind every answer?
- What happens when failure codes are inconsistent in our history?
- Does any maintenance data leave our network, including for model hosting?
- Can the assistant ever close, release or change a work order without approval?
- How do you evaluate answer quality before we rely on it?
- What does IFS already offer in this area that we should use first?
- What is the ongoing cost in GPUs and support once the pilot ends?
Frequently asked questions
Can AI answer questions about IFS Cloud work orders?
Yes. A layer that reads IFS Cloud through its APIs can answer questions across work orders, tasks, objects, parts and costs in plain English, show the query used, and respect the asker's permissions. Answer quality depends on how consistently work and failure data is recorded.
Does it replace IFS maintenance planning?
No. Planning boards, scheduling, approvals and work execution stay in IFS. The AI layer answers questions and drafts text. It works best as an addition for cross-object analysis and search that standard screens handle slowly.
Is IFS's own AI enough?
IFS offers its own AI features within the product, and for many standard scenarios they may be enough. A separate layer makes sense when you need models on your own infrastructure, cross-system questions, or retrieval over your own documents.
How is repeat failure analysis made reliable?
By grounding it in structured data such as failure codes and object IDs, and using text search only to supplement. The answer shows the records behind it. If failure codes are poor, a short cleanup improves results more than any model change.
Can it run without internet access?
Yes. Open-weight models, the index and orchestration can run entirely inside your network. IFS connectivity depends on where IFS is hosted, which is worth confirming at the start.
Can technicians use it on the shop floor?
Yes, through a browser or mobile interface behind your identity provider, with the same IFS permissions. Keep answers short and cite the source work orders so technicians can verify them.
How long does a pilot take?
A read-only pilot on one site typically takes a few weeks after access is granted. Most of the time goes into data checks and tuning against real planner questions, not into model setup.
Related guides
AI for IFS Cloud in aerospace and defense manufacturing
Add AI to IFS Cloud for aerospace and defense: projections, Aurena UX, Lobby, and engineer-to-order data, deployed on-prem or in your private cloud.
Maintenix + on-prem AIAI for IFS Maintenix in aviation MRO
AI over IFS Maintenix maintenance records, AD/SB compliance, and task cards for MRO, deployed on-prem so tech records never leave your network.
IFS FSM + private AIIFS Field Service Management AI Integration: Private AI on IFS Cloud Service Data
AI for IFS Field Service Management: service requests, work tasks, agreements and technician data via IFS Cloud APIs, joined to parts and finance. On-prem.
MRO technical records + on-prem AIAI for aviation MRO maintenance records and technical documentation
On-prem AI reads AD and SB packets, cross-checks logbooks and work orders against your MRO ERP, and drafts compliance narratives, without technical records leaving your network.
Agents + approval gatesAI Agents for ERP, Running On-Prem
A practical guide to on-prem AI agents for ERP: what they can safely automate, where human approval belongs, and how to design the guardrails.
90-day ERP AI pilotA 90-Day ERP AI Pilot Plan With Real Success Criteria
A week-by-week 90-day plan for piloting AI on your ERP, with defined success criteria for each phase so the pilot ends in a clear go or no-go decision.
Plan it with numbers
On-Prem AI ROI Calculator
Turn hours saved per employee into annual net benefit, payback months, and 3-year ROI for an on-prem AI investment.
Free ToolOn-Prem AI Deployment Checklist
A 30-point pre-deployment checklist covering use cases, hardware, security, model operations, and rollout for self-hosted enterprise LLMs.
Free ToolOpen-Weight Model Selector
A 10-question assessment that matches your hardware budget, workload complexity, and operational maturity to the right open-weight model size class.
GuideSyteLine MRP Tuning & Optimization
Tune SyteLine MRP parameters with AI for optimal inventory levels. Reduce excess stock 25-40% while improving fill rates in Infor CloudSuite Industrial.
GuideSyteLine Production Scheduling with AI
Transform SyteLine production scheduling with AI. Automated APS optimization, finite capacity scheduling, and real-time shop floor synchronization.
GuideSyteLine BOM Management & Optimization with AI
Optimize SyteLine Bill of Materials with AI. Automated BOM validation, cost rollups, and engineering change management for Infor CloudSuite Industrial.
Talk it through with an engineer who knows IFS Cloud
Bring one real question your team cannot answer from the ERP today. We will map the data path, the model, and where it runs, and tell you honestly if AI is the wrong tool for it.