Epicor 9 and Vantage Private AI Layer Blueprint Generator
This generator produces a private AI architecture plan for a legacy Epicor 9 or Vantage system on Progress OpenEdge. It assumes read-only access through the OpenEdge SQL engine and a replicated reporting copy.
Fill in
Epicor 9 and Vantage Private AI Layer Blueprint Generator
Private AI Layer Architecture Plan: Epicor 9 and Vantage
Prepared for: Lakeshore Machining Co System: Epicor 9 and Vantage Epicor 9.05.7xx on Progress OpenEdge 10.2B Hosting: Air-gapped network Named users: 70 across 1 site(s) Target go-live: 2027-04-01
1. Context and goals
This plan describes a private AI layer over a legacy Epicor 9 or Vantage system on Progress OpenEdge. The legacy database is treated as read-only and is replicated to a reporting copy so that analytical and model-driven queries never run against the live system.
Main use cases: Open orders and job status Inventory and shortages Purchase order follow-up Historical sales lookups
Data that must stay on site: Customer drawings and contract terms Costing and rate data
GPU position: None yet.
2. Components
| Component | Design for Lakeshore Machining Co | Notes |
|---|---|---|
| Data access path | OpenEdge SQL-92 engine over ODBC or JDBC with a read-only user; replication or scheduled extract to a SQL Server or PostgreSQL reporting copy | Check SQL width settings so long text fields are not truncated. |
| Retrieval index | Local vector store over documents on file shares plus text fields extracted from the reporting copy | Every chunk carries source, site and access tags; untagged content is excluded. |
| Model serving | Open-weight model on vLLM or Ollama, sized for 70 named users on None yet | Model, quantisation and version recorded in the audit log. |
| Tool server | Tool server with parameterised read actions against the reporting copy; no direct database credentials in the model path | Parameterised read actions only; row caps and column allow lists. |
| Identity and permissions | Directory authentication mapped to a maintained table of users, companies and plants, because permissions live in the application and not in SQL | Unknown users and unmapped content default to no access. |
| Audit | Append-only log of user, prompt, tool call, parameters, row counts and answer | Retained per Lakeshore Machining Co records policy; reviewed at each rollout phase. |
3. Security controls mapped to your inputs
| Control | Implementation | Driven by |
|---|---|---|
| Data residency | Replica, index and model all inside the Air-gapped network boundary; no outbound API calls at run time | Data that must never leave the site |
| Least privilege | Read-only OpenEdge SQL user; reporting copy accessed by a separate read-only login | System version and role model |
| Network isolation | Dedicated segment for model and tool servers; only the tool server reaches the data source | Hosting |
| Prompt injection defence | Model receives bounded tool results only; no direct database or file access | Use cases |
| Write-back | No write-back in scope; if later required, only through Epicor's supported service layer after human approval | Phase 3 scope |
4. Rollout plan
| Phase | Scope | Exit criteria |
|---|---|---|
| 1: Read-only (weeks 1-6) | Approved read actions and document Q&A for a pilot group at one site | Accuracy review of sampled answers; no permission leaks in adversarial tests |
| 2: Drafts (weeks 7-12) | Summaries, reports and draft documents saved for human completion | Users edit rather than rewrite; audit log reviewed by IT and data owners |
| 3: Approved actions (week 13 onward) | Narrow, reversible transactions via supported interfaces after explicit approval | Rollback tested per action type; owner sign-off per action |
5. Open questions for IT
Which OpenEdge version and SQL broker configuration are in use? Is SQL width set appropriately for text columns? Which replication option will be used (OpenEdge replication, Pro2 or scheduled extract)? How current must the reporting copy be? Which customisations change the numbers shown on screens? Is a migration planned that the data mapping should support?
Approvals
Starting point for discussion, not a design guarantee. Table names, driver versions and SQL behaviour depend on your OpenEdge and Epicor release; verify against your data dictionary.
Epicor 9 and Vantage architecture review
Optional: have an engineer review your generated blueprint against your environment and flag gaps before you buy hardware.
Optional. The tool and every export work without this. No spam.
How to use it
Select your Epicor and database versions, the hosting model, user and site counts, and your GPU position. Write the use cases as questions people ask today, and list the data that must stay on site. Many legacy sites are already isolated, so air-gapped is preselected; change it if your network differs.
- Confirm the OpenEdge SQL broker is running
- Name who owns the data dictionary
- State how fresh the reporting copy must be
- List customisations that alter reported figures
What the plan contains
Context, a component table covering the SQL read path, retrieval index, model serving, tool server, identity and audit, controls mapped to your inputs, a three-phase rollout, open questions for IT, and approvals. Write-back is out of scope by default for a legacy release.
Why replicate
Progress OpenEdge handles SQL and 4GL workloads on the same database, and heavy SQL can slow the application. A replicated copy lets you add indexes, flatten tables and keep history, and lets the model run broad queries safely. It also gives a clean source for data mapping if a migration follows.
- OpenEdge replication or Pro2 style replication
- Scheduled extracts for smaller sites
- Views that flatten joins for the assistant
Common mistakes
Frequent problems are truncated text from SQL width, heavy queries on the live system, and trusting figures that depend on customisation. The plan makes the replica, the permission mapping and the show-the-query rule explicit.
Frequently Asked Questions
Can this run on a very old Epicor version?
Often yes, because it needs only read access through OpenEdge SQL and a reporting copy. Driver, table and SQL width details vary by release, so verify against your data dictionary and test with a single query first.
Why is write-back excluded?
Writes to a Progress database outside the application bypass business rules and are risky. If writes become necessary, use Epicor's supported service layer after human approval, ideally after any planned migration.
Does the plan help with a migration?
Indirectly. The replica, views and documented data model are the same assets a migration needs for mapping and validation, so the work is not wasted if you move to Kinetic or another ERP.
What if we have no GPU?
Select None yet and use the sizing calculator to estimate what to buy. A small pilot can run on one 24 GB GPU, and the plan assumes you validate throughput before committing to larger hardware.
If you want an engineer to review the plan against your Epicor 9 environment or build it on site, Netray can help. The generator is free without contact.
Related Tools
Epicor 9 and Vantage AI Assistant Sizing and Value Calculator
Estimate GPUs, index size, hours saved and payback for a private AI assistant over a legacy Epicor 9 or Vantage system.
On-Prem AIGPU Sizing Calculator for LLM Inference
Work out how many GPUs you need to serve a given open-weight model to your user base, based on memory footprint and token throughput.
On-Prem AIRAG Infrastructure Sizing Calculator
Estimate vector storage, node RAM, generation GPUs, and monthly infrastructure cost for a retrieval-augmented generation deployment over your document corpus.
On-Prem AIPrivate AI Total Cost of Ownership Calculator
Model the full 3-year cost of an on-prem AI deployment, including amortized hardware, power, staff time, and support, against comparable API spend.
More Epicor tools
See all- Epicor Kinetic Job Status Reference
- Epicor BAQ Query Template Generator
- Epicor BPM Directive Design Template
- Epicor DMT Import Template Worksheet (Part and BOM)
- Epicor MRP Run Checklist
- Epicor Kinetic Upgrade Checklist
- Epicor REST API v2 URL Builder
- Epicor Security Group Design Worksheet
- Epicor Kinetic Customization Change Request Form Template
Keep exploring
Free tools
- Epicor ERP 10 Private AI Layer Blueprint GeneratorGenerate a private AI architecture plan for Epicor ERP 10 on-prem, covering REST and BAQ access, replica, retrieval, local model serving, security and rollout.
- Epicor 9 and Vantage AI Assistant Sizing and Value CalculatorEstimate GPUs, index size, hours saved and payback for a private AI assistant over a legacy Epicor 9 or Vantage system.
- Epicor Kinetic Private AI Layer Blueprint GeneratorGenerate a shareable private AI architecture plan for Epicor Kinetic on-premises, with components, security controls, rollout phases and open questions for IT.
ERP fixes and how-tos
- AI for Epicor 9 and Vantage on Progress OpenEdge: A Private ArchitectureAdd a private AI assistant to Epicor 9 or Vantage on Progress OpenEdge: SQL-92 read path, replication, local LLM, permissions and honest limits.
- How to Add AI to Epicor Kinetic On-Premises Without Moving Your DataReference architecture for a private AI assistant on Epicor Kinetic on-premises: BAQ and REST v2 access, SQL replica, local LLM, roles, audit and rollout.
- Epicor ERP 10 AI Integration Without Cloud: Private ArchitectureAdd a private AI assistant to Epicor ERP 10 on-prem: REST and BAQ access, SQL replica, Service Connect, local LLM, role mirroring, audit and phased rollout.
Integrations
- How to add AI agents to Epicor Vantage and Epicor 9How to add AI agents to Epicor Vantage or Epicor 9: read from the Progress database, act through business objects, keep approvals human, and run it on-prem.
- How to connect Power BI to Epicor Vantage and Epicor 9Connect Power BI to Epicor Vantage or Epicor 9 through the Progress ODBC driver, an on-prem gateway or a warehouse, with manufacturing dashboards and AI Q&A.