Manufacturing SystemsERP Platform

HxGN EAM + private AI

AI for HxGN EAM, Formerly Infor EAM: Ask Your Asset Records Questions Without Exporting Them

Short answer

AI for HxGN EAM connects a private language model to the EAM web services and database so asset managers, planners and technicians can ask plain-English questions about equipment, work orders, stores and purchasing and get answers grounded in live records. It runs on-prem or in a sovereign cloud, and fits sites that came from Infor EAM or Datastream 7i and still run the same data model.

ERP: HxGN EAM, Infor EAM
Industries: Utilities, Manufacturing, Aerospace, Defense
Written for: Asset Management Director

Discuss your HxGN EAM environment See use cases

HxGN EAM is the product Hexagon took over from Infor, where it was Infor EAM and, before that, Datastream 7i. Many customers still call it Infor EAM, still run it next to Infor ERP products through ION or other integrations, and still see the original R5 table naming in reports and integrations. It is heavily used in utilities, mining, manufacturing, airports and defense-related facilities, where asset lifecycles are long and records outlive staff.

The strength of the system is its depth: equipment hierarchies, work orders, PM schedules, inspections, stores and purchasing, labor and contractors, all configurable. The cost of depth is that finding a particular answer means knowing which screen, grid and filter to use. Experienced users build and share grids, but new planners and managers rely on those few people or wait for custom reports.

If you are on Hexagon's current cloud offering, check its own newer capabilities first. A private AI layer is worth considering when you run an older on-prem version for years to come, when asset data is subject to security or contract limits on cloud AI, or when your questions must combine EAM records with an ERP, drawings and documents. This is especially relevant for Infor shops that already use SyteLine, M3 or LN and want one place to ask across maintenance and operations.

This page describes how that layer connects, what is realistic to start with, and the questions to settle before connecting AI to a system that manages critical assets.

What usually gets in the way

The problems we hear most from asset management director teams running HxGN EAM.

Knowing which grid and filter to use

EAM screens are powerful but specific. Users who do not know the right grid or the right user-defined fields struggle to answer questions that a power user can answer in minutes.

Long asset lifecycles outlive the people who understand them

Equipment may be decades old, with histories that span several system migrations and many technicians. Context about why a decision was made is in long text fields and attachments.

Mixed Infor and Hexagon vintages

Customers often run a mix of versions, rebrandings and ERP integrations, and documentation or consultants refer to different names for the same screens and objects.

Stores and purchasing integration gaps

Parts, requisitions and purchase orders may span EAM and ERP, so determining what is on order and for which work order requires looking in two places.

Security and contract limits on data handling

Utilities, defense-adjacent sites and critical infrastructure often restrict where asset and location data may be processed, ruling out public AI tools.

Where AI earns its place in HxGN EAM

Each use case names the ERP objects it reads or writes, so your ERP team can judge the integration effort before anyone commits budget.

Natural-language questions across equipment and work orders

Planners ask which open work orders on a given equipment class are overdue or waiting on parts, with the filter shown.

Touches: Equipment (R5OBJECTS), Work Orders (R5EVENTS), Parts (R5PARTS), Stock (R5STOCK)

Outcome: Reduces dependence on a few power users for ad hoc questions.

Equipment history briefing

Produces a short summary of an asset's work history, repeat failures, major repairs and attachments before a repair-or-replace review.

Touches: Equipment, Work Orders, comments and long text, Documents

Outcome: Cuts preparation time for asset review meetings, with each statement linked to a record.

PM and inspection review

Flags PM schedules and inspections that rarely find anything or are repeatedly deferred, as a list for engineers.

Touches: PM schedules, Work Orders, Inspections, Equipment

Outcome: Produces an evidence-based list for program tuning.

Stores and purchasing visibility

Shows parts below minimum, parts on order for open work orders and slow-moving stock, optionally joined to ERP supplier data.

Touches: Stores, Parts, Requisitions, Purchase Orders, ERP item and vendor records

Outcome: Highlights stockout risk and obsolete stock without a custom report each time.

Contractor and labor questions

Answers questions on hours by craft, contractor use and completion rates by area.

Touches: Labor bookings, Work Orders, Crafts, Departments

Outcome: Supports resource planning conversations with numbers on hand.

Cross-system queries with the ERP

Combines maintenance cost from EAM with ERP cost centers and fixed asset values to rank assets by total cost.

Touches: Equipment, Work Orders, ERP cost center and fixed asset records

Outcome: Joins two views that today are reconciled by hand.

Manual and drawing retrieval

Retrieval over manuals, drawings metadata and procedures, cited to the asset or work order in question.

Touches: Documents, Equipment, Work Orders, external document stores

Outcome: Reduces time spent hunting for the right page or revision.

Reference architecture

The layer connects to HxGN EAM through its web services and, where allowed, a read replica of the database, keeps a private index with ERP and documents, and serves answers with a local model. Anything written is a draft approved by a person.

1
EAM connector
Uses the HxGN EAM web services (SOAP and REST, including grid-based data access) with a dedicated integration user, so queries run under EAM's own security. For large analytical questions, a read replica of the EAM database can be used with agreed table access.
2
Semantic layer
Maps EAM entities, user-defined fields, classes and statuses to plant language, bridging old R5 naming and current product terms and aligning equipment codes with ERP assets.
3
Model serving
Open-weight models served with vLLM or Ollama on customer GPUs or in a sovereign cloud, separate from EAM application servers.
4
Retrieval and agents
Structured queries for facts, vector retrieval over long text, manuals and procedures, and drafting agents held behind approval.
5
Governance and audit
Per-user logs of question, query and answer, mapped to EAM security and to organization or department restrictions.

Integration notes for your ERP team

Use a dedicated EAM integration user with a narrow user group, scoped to the organizations and departments in scope.
Prefer web services and grids, which apply EAM's own security, over direct table access, and reserve a read replica for heavy analytics.
Account for user-defined fields and custom screens, which are common and differ between sites.
Align equipment codes with ERP asset numbers; for Infor ERP customers, check how ION or existing integrations already map them.
Index long text, comments and attachments with equipment and work order metadata so retrieval can be filtered.
Keep write-back to drafts a person approves, created via the web services so EAM business rules and workflows still run.
Test against your exact version, since older Infor EAM and current HxGN EAM releases differ in interfaces and naming.

Deployment options

Air-gapped on-prem

Utilities, defense-adjacent facilities and critical infrastructure with strict data handling.

Connector, index and model run inside the network that hosts EAM, with no outbound path.

Sovereign or private cloud

Customers on Hexagon's cloud or a managed environment who need AI under their own tenancy and region.

A separate tenant hosts the store, index and model, reading EAM over secured web services.

Hybrid with Infor ERP

Sites running HxGN EAM next to SyteLine, M3 or LN.

One layer sits next to the ERP and reads EAM over web services, so maintenance and operations questions are answered together.

Compliance and data control

How the architecture supports your obligations. Certification and accountability stay with your organisation; the design keeps the evidence straightforward.

Critical infrastructure security (NERC CIP for bulk electric systems, NIS2 in the EU)

A private, access-controlled deployment lets the customer apply its own electronic access and monitoring controls; scoping which asset data the assistant may index is the customer's decision.

ITAR and export-controlled technical data

Running inside the customer's boundary lets them keep controlled data on infrastructure they already govern, with access limited to authorized persons.

ISO 55001 asset management

Evidence-based review lists and citations support documented asset decisions, which remain with accountable people.

OSHA 1910.147 and site permit-to-work procedures

Retrieval cites the current approved procedure and does not replace the permit or isolation process.

GDPR and regional data protection

Labor and contractor personal data can be excluded or pseudonymized in the index.

Where Netray fits

Custom build

EAM installs are heavily configured and span several versions, so a bespoke connector and semantic layer built around your fields and security model is the safe approach.

ERPray

For Infor ERP sites, ERPray covers the ERP side with read-only, role-aware answers, and can sit alongside an EAM connector for cross-system questions.

DataRay

Useful when manuals, drawings and shared drives need to be searched together with EAM data on-prem.

How an engagement runs

Phase 1 . 2-4 weeks

Discovery

-Version, interface and customization review
-Security model and data boundary plan
-Use cases ranked by planner and asset manager pain
-ERP mapping approach

Phase 2 . 6-8 weeks

Pilot

-Read-only connector against a test environment
-3 use cases for one organization or site
-Accuracy review against trusted grids and reports
-Logging validated against EAM security

Phase 3 . 4-6 weeks

Production

-Production service account and monitoring
-Audit logging
-Runbook for new use cases
-User training

Phase 4 . Ongoing

Scale

-More organizations and sites
-ERP and document sources
-Approved drafting flows
-Quarterly accuracy review

Questions to ask any vendor, including us

A short list that separates real HxGN EAM AI work from a chatbot demo.

Which EAM interface and version will the connector target, and how will it handle our user-defined fields?
Does access go through EAM's web services and security, or direct database credentials?
How are organization and department restrictions enforced per question?
How do you reconcile equipment codes with our ERP?
Where do the model and index run, and what is the network path?
Can every answer cite the underlying records?
When would we be better off using Hexagon's own newer capabilities?

Frequently asked questions

Is HxGN EAM the same as Infor EAM?

HxGN EAM is the Hexagon product that grew out of Infor EAM, and before that Datastream 7i. Many sites still use the Infor EAM name, and older data structures such as R5-prefixed tables remain relevant. The connector is built for the version you actually run.

Can this run on-prem?

Yes. The connector, index and model can run inside the network that hosts EAM with no outbound path, which suits utilities, defense-adjacent sites and other restricted environments.

Do you read the database directly?

Preferably through the EAM web services and grids, so EAM's own security applies. A read replica may be used for heavy analytics where policy allows, with table access agreed in advance.

Can it work with our Infor ERP?

Yes. For SyteLine, M3 or LN sites, the layer can read the ERP read-only and EAM together, so questions about maintenance cost, parts on order or asset value cross both systems. Quality of the equipment mapping decides reliability.

Will it create work orders?

It drafts them for approval, and approved records are created via the web services so EAM business rules and workflows run as usual. Nothing is written without a person approving.

What about Hexagon's own AI features?

If you are on the current cloud product, evaluate those first since they are integrated and supported. A private layer is for older on-prem versions, restricted environments, or questions across several systems and documents.

How long does a first deployment take?

Usually discovery of a few weeks and a pilot of six to eight weeks for one site, depending on customizations, access arrangements and the number of sources. Larger rollouts follow once answers are trusted.

Related guides

IBM Maximo + private AI

AI for IBM Maximo That Reads Your Work Orders and Failure History Without Leaving Your Boundary

Add grounded, private AI to IBM Maximo Manage or 7.6: ask questions over work orders, assets and failure codes via the REST API, on-prem or sovereign cloud.

Maintenix + on-prem AI

AI 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.

SAP PM/EAM + on-prem AI

AI for SAP Plant Maintenance: Notifications, Work Orders, and Predictive Signals

AI grounded on SAP PM data that triages notifications, drafts work orders, standardizes failure codes, and flags predictive maintenance signals, on-prem.

RAG + SQL + permissions

A Private LLM Grounded on Your ERP Data

How a private LLM answers questions on your ERP data: RAG plus text-to-SQL, role-based permissions inherited from the ERP, and where each fits.

Agents + approval gates

AI 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.

Opcenter Execution + private AI

AI for Siemens Opcenter Execution MES, Running Inside Your Own Network

Add private, grounded AI to Siemens Opcenter Execution MES: ask questions over lots, work orders, nonconformances and genealogy without leaving your network.

Plan it with numbers

Free Tool

Talk it through with an engineer who knows HxGN EAM

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.

Start the conversation See ERPray

All AI for ERP guides