Propel PLM + private AI
AI for Propel PLM: Using Salesforce Data Access Properly, With Models You Control
Short answer
Propel PLM runs on the Salesforce platform, so AI for Propel reads item, revision, BOM and change records through Salesforce APIs under the asking user's sharing and field-level security. A privately hosted model then answers questions and drafts change text with citations. It is useful where Salesforce's own AI features do not reach product data or where you need to control hosting.
- ERP
- Propel PLM
- Industries
- Medical Devices, Consumer Products, Electronics, Life Sciences
- Written for
- Product Operations Manager
Propel PLM is built as a managed application on Salesforce, so items, revisions, BOMs and change orders are Salesforce records, governed by the same profiles, permission sets, sharing rules and field-level security as the rest of your org. For companies already living in Salesforce for sales and service, that puts product data next to customer data, which is its appeal.
That architecture changes what AI integration looks like. Data access is via the Salesforce REST API, SOQL and Bulk API with OAuth through a connected app, and automation can run in Apex or Flow. It also means Salesforce's own AI offerings are a natural first look, and an honest assessment starts by asking whether they cover your needs.
They often cover CRM-adjacent tasks well. Where they fall short for product teams is in cross-system questions that combine PLM, ERP and quality data, in regulated settings that restrict where inference runs, and in customer-specific change-control rules that require predictable, citation-based behavior.
This page covers a practical private design for Propel: how to read the data safely, how to keep sharing rules in force, where AI helps change, quality and launch teams, and where Salesforce-native tools or standard reports are enough.
What usually gets in the way
The problems we hear most from product operations manager teams running Propel PLM.
Product data and ERP data live apart
Item and BOM data sits in Salesforce while costs, stock and supplier orders sit in the ERP. Cross-questions need exports and manual joins.
Change orders are written inconsistently
Change justification and impact text vary by author, making review slower and audit trails harder to follow for regulated products.
Report building is limited by skills
Salesforce reports and dashboards are powerful, but questions across item revisions, BOM lines and changes need someone who knows the object relationships.
Governor limits and data volume
Large BOM explosions and bulk reads run into API and processing limits, so naive integrations are slow or fail.
Security model must stay intact
Sharing rules, permission sets and field-level security protect sensitive costs and supplier terms. Extracting data with an over-privileged integration user undermines them.
Where AI earns its place in Propel PLM
Each use case names the ERP objects it reads or writes, so your ERP team can judge the integration effort before anyone commits budget.
Ask-your-PLM for items and BOMs
Answer where-used, BOM revision and lifecycle questions in plain English, returning item records with links back into Propel.
Touches: Propel item, item revision and BOM records, lifecycle phase fields
Outcome: Routine structure lookups no longer need a report builder or an administrator.
Change order drafting
Draft reason for change, affected item lists and impact statements from selected records and past similar changes, for author review.
Touches: Propel change orders, affected items, attachments
Outcome: More consistent change records and quicker reviews.
Quality event linkage
Connect complaints, nonconformances or CAPA records to affected items and changes so recurring failure modes surface.
Touches: Propel quality records, item revisions, related changes, Salesforce case data where used
Outcome: Trends across product, change and service data become visible with less manual correlation.
BOM to ERP comparison
Compare released Propel BOMs against ERP manufacturing BOMs and list quantity, unit and revision differences.
Touches: Propel BOM records, ERP BOM and item master, integration logs
Outcome: Mismatches found at release, not during production planning.
New product launch readiness
Check item completeness, compliance attributes, supplier approvals and open changes against your launch checklist.
Touches: Item fields, supplier records, compliance and document attachments, task records
Outcome: Fewer gaps at launch gates; the checklist remains yours.
Document retrieval
Answer from controlled documents and specification attachments, citing the file and version.
Touches: Content versions, attached files, document records
Outcome: Quicker answers during audits and customer reviews.
Cross-system executive questions
Combine PLM revision status, ERP inventory and supplier lead times to answer questions such as which new revisions are blocked by missing components.
Touches: Propel items and changes, ERP on-hand, open POs, vendor lead times
Outcome: Leaders get cross-system answers without a manual data-gathering round.
Reference architecture
A connector authenticates to Salesforce through a connected app, queries Propel objects with SOQL and the REST API as the asking user, and hands retrieved records to a privately hosted model. Drafts are created as records for human review.
- 1
Authentication
OAuth through a connected app with user-delegated flows for questions and a tightly scoped integration user for indexing.
- 2
Data access
SOQL via REST API for interactive questions; Bulk API for scheduled indexing; field-level security and sharing evaluated by Salesforce.
- 3
Retrieval layer
Object-aware structured lookup for items, revisions and BOMs, plus a vector index over attachments, preserving record ids for citation.
- 4
Model layer
Open-weight models running on your GPUs or private cloud, with no calls to public inference APIs.
- 5
Governance layer
Per-user logs, limit monitoring, answer evaluation on known cases, and draft-only writes.
Integration notes for your ERP team
- Use a connected app with scoped OAuth permissions; avoid a system administrator profile for the integration user.
- Check Propel's object and field names against your installed package version; managed package namespaces and field sets differ across releases.
- Use Bulk API for scheduled indexing and REST with SOQL for interactive queries, and monitor daily API request limits.
- Evaluate sharing as the asking user; do not cache query results across users without keeping their permissions attached.
- Handle large BOMs by paging and by querying levels rather than whole explosions in one call.
- Read ERP data through its own API for comparisons so the check does not depend on the synchronization under review.
Deployment options
Private cloud inference
Most Propel customers, who already run Salesforce in the vendor cloud
The retrieval and model services run in your own cloud tenancy and call Salesforce APIs over private or restricted network paths.
On-premises GPU
Companies with strict data handling for supplier terms or regulated product data
Extracted data and models are kept on-site. Salesforce stays vendor hosted, so the design covers only data after extraction.
Salesforce-native first
Teams whose needs are CRM-adjacent and non-sensitive
Start with Salesforce's own AI features and add a private layer only for the gaps. This can be the cheapest path and should be evaluated honestly.
Compliance and data control
How the architecture supports your obligations. Certification and accountability stay with your organisation; the design keeps the evidence straightforward.
FDA 21 CFR Part 820 (QMSR) and ISO 13485
Change and design records are read only. Draft text is labeled and must be approved through Propel's own workflow, so the controlled record is unchanged.
21 CFR Part 11
Electronic signatures remain in Propel. The AI layer never signs or approves anything.
EU MDR and UDI traceability
Item and document references are cited exactly. Regulatory submissions still rely on your regulatory team's review.
SOC 2 and customer security reviews
Because data is copied out of Salesforce, document where it lands, who can reach it, retention and encryption as part of your own control environment.
Where Netray fits
Custom build
A Salesforce-aware connector, change-drafting agent and BOM comparison tailored to your Propel configuration.
ERPray
Covers the ERP side for cross-system questions with read-only queries and visible SQL.
DataRay
Useful when attachments, shared drives and ERP tables must sit behind one private chat.
How an engagement runs
Phase 1 . 1 to 2 weeks
Discovery
- -Review of Propel configuration and object model
- -Comparison against Salesforce-native AI options
- -Security and data-handling decision
- -Question inventory
Phase 2 . 3 to 5 weeks
Pilot build
- -Connected app and delegated access
- -Item, BOM and change retrieval
- -Cited answers
- -Evaluation set
Phase 3 . 3 to 4 weeks
Workflow agents
- -Change drafting
- -ERP BOM comparison
- -Launch readiness checks
- -Quality linkage report
Phase 4 . 2 weeks
Handover
- -Runbook and limit monitoring
- -Credential rotation procedure
- -Model update plan
- -User training
Questions to ask any vendor, including us
A short list that separates real Propel PLM AI work from a chatbot demo.
- Have you compared this with Salesforce's own AI features for our use cases?
- Do queries respect sharing rules and field-level security for each user?
- What does the integration user have access to, and how is it scoped?
- Where does extracted data sit, and for how long?
- How do you stay within Salesforce API limits at our volume?
- How do you handle package upgrades that rename or add fields?
- What is written back to Propel, and who approves it?
- How do you test answers on our actual BOMs?
Frequently asked questions
Should we just use Salesforce's own AI?
Evaluate it first. For CRM-adjacent summaries and simple record questions it may be enough. A private layer makes sense when you need your own model hosting, cross-system answers with ERP and quality data, or tightly controlled, citation-based change drafting.
Does the AI respect Salesforce sharing and field-level security?
It should, by querying as the asking user through delegated OAuth so Salesforce evaluates access. An integration user with broad access answering for everyone is a design to avoid, particularly for cost and supplier fields.
Can it modify Propel records?
By default, no. It reads and drafts. Changes are created and approved by users in Propel, retaining the audit trail and signatures required for regulated products.
Will Salesforce API limits be a problem?
They can be, for large BOMs or frequent syncs. Incremental indexing, Bulk API for batches, caching of non-sensitive structure and paged queries keep usage within limits, and monitoring flags trouble early.
Is on-prem possible given Salesforce is cloud?
The AI stack can be on-prem or in your private cloud, pulling data over APIs. Propel and Salesforce remain vendor hosted, so the privacy benefit applies to inference and the copied dataset, not to the source system.
What about our ERP?
ERP data is read separately through its own API or read-only views. This lets the AI compare Propel BOMs against ERP BOMs and answer questions mixing revision status with stock and purchasing.
How long does a pilot take?
Often four to eight weeks from access to a working read-only pilot, depending on how customized the Propel configuration is and how quickly connected app and security approvals are granted.
Related guides
AI for Engineering Change Impact on BOMs, Routings, and Open Orders
AI reads a proposed ECO against live BOM, routing, and open order data to show engineering exactly what a change touches before it gets released.
Faster paperwork, same standardAI for NCR, CAPA, and 8D Drafting in AS9100 and ISO 9001 Quality Systems
AI drafts NCR, CAPA, and 8D reports from ERP quality data for engineer review, cutting drafting time under AS9100 and ISO 9001 without changing sign-off.
EMS / PCBA + on-prem AIAI for EMS and PCBA Manufacturers Running SyteLine, Epicor, or NetSuite
AI for EMS and PCBA manufacturers on SyteLine, Epicor, or NetSuite: automate BOM scrubbing, AVL checks, and obsolescence alerts, grounded in your own ERP data on-prem.
RAG + SQL + permissionsA 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.
Ask your ERP anythingNatural Language Query for ERP Data: Ask SAP, Infor, or Oracle a Question in Plain English
See how natural language query over SAP, Infor, Oracle, and NetSuite data works: grounded text-to-SQL, role-based permissions, and a visible audit trail.
Vendor copilots vs private AIBuild vs Buy: Should You Use Your ERP Vendor's AI Copilot, or Build Your Own?
SAP Joule, Copilot for D365, Infor GenAI, Oracle AI Agent Studio, or a private LLM on your own data. A CIO framework for the build vs buy decision, with real trade-offs.
Plan it with numbers
ECN Impact Assessment
Score how well your engineering change process controls WIP, inventory, effectivity, and downstream impact before changes hit the floor.
Free ToolElectronics BOM Cost Calculator
Estimate loaded BOM cost per board and annual material spend, including volume pricing tiers, procurement overhead, and attrition.
Free ToolLLM API vs Self-Hosted Cost Calculator
Model your API bill from requests and token mix, compare it against an all-in self-hosted monthly cost, and see monthly and annual savings.
GuideERP-PLM Integration Guide
ERP-PLM integration guide: sync engineering BOMs, item masters, ECOs, and CAD documents between PLM and Infor SyteLine or LN without duplicate data entry.
GuideAI Change Impact Analysis for ERP Systems
Use AI agents for ERP change impact analysis predicting downstream effects of configuration changes, patches, and customizations before deployment.
GuideERP-CAD Integration for Engineering Data
ERP-CAD integration for engineering data: push SolidWorks, Creo, and Inventor BOMs and drawings into ERP item and routing records with controlled revisions.
Talk it through with an engineer who knows Propel PLM
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.