Before adding an AI agent, make sure EnterpriseOne is ready to be called deliberately.
Assess whether a JD Edwards EnterpriseOne estate is ready for agentic AI across identity, AIS, Orchestrations, metadata, security and data residency.
AI readiness is not mainly about model selection. The hard questions are whether identity is trustworthy, whether useful actions already have bounded interfaces, and whether the AI can discover JDE facts without hallucinating them.
Readiness checklist
| Area | Question |
|---|---|
| Identity | Can the user authenticate through JDE, enterprise SSO, Windows or a trusted IdP? |
| AIS | Are AIS endpoints reachable and appropriately secured? |
| Orchestrations | Are useful business operations already encapsulated and permissioned? |
| Metadata | Can the AI obtain authoritative table, data-item and key information? |
| MCP | Which tools should be exposed, to which clients and groups? |
| Data residency | Which content may go to hosted models and which must remain local? |
Start with one bounded use case
A good first agent is not “do anything in JDE.” It is a small collection of high-value, explainable capabilities with known inputs, outputs and security rules.
Continue into the wider architecture
For a broader reference model, JDE Citizen Developer's Enterprise AI Integration Patterns shows how identity, MCP, policy, local tools and security platforms fit together. The implementation products themselves are on Beanstalk.