Oracle JD Edwards and Zendesk integration

What moves between them

Data flows one direction: from Oracle JD Edwards into Zendesk. ml-connector polls JD Edwards suppliers and organizations on a daily schedule, mapping them to Zendesk organizations so that support tickets tagged with a company automatically surface the correct ERP supplier record. Separately, it polls open invoices and purchase orders tied to a supplier and indexes them by invoice number, PO number, and supplier name, making them searchable within Zendesk ticket comments or custom fields. Because Zendesk is a ticketing system with no GL posting capability, the flow is read-only: ml-connector enriches tickets with ERP context but never writes financial data back to JD Edwards.

How ml-connector handles it

ml-connector stores the JD Edwards AIS Server URL and credentials encrypted, obtains a session token on each poll cycle, and queries the Address Book, Accounts Payable Ledger, and Purchase Order tables with a date filter on the last update timestamp. It maps JD Edwards suppliers to Zendesk organizations by name or external ID, and stores open invoices and orders in a searchable index keyed by supplier, invoice number, and PO number. When a support ticket mentions a supplier name or invoice number, agents can reference the enriched data without leaving Zendesk. JD Edwards session tokens expire after 30 to 60 minutes, and ml-connector re-authenticates on the next poll cycle; it also respects the AIS Server rate limits and pagination boundaries (maxPageSize and moreRecords continuation). Because JD Edwards is on-premises, the customer must whitelist ml-connector's egress IP address in their AIS Server allowlist, and the connector validates that the service account license remains valid on each token refresh.

A real-world example

A mid-sized manufacturing company runs Oracle JD Edwards for supply chain and finance, and uses Zendesk for customer support. When a customer calls with a question about an invoice or delivery delay, the support agent previously had to ask the customer for an invoice number, log into JD Edwards separately to look up the PO and shipment status, then log back into Zendesk to document the findings. With the integration, the agent searches the ticket for the supplier name or invoice number, and Zendesk instantly surfaces the open invoice, the original PO line items, and the expected delivery date from JD Edwards, all within the support interface. Supplier-related issues are resolved faster, and the support handoff to the ops team is more complete.

What you can do

• •Poll Oracle JD Edwards suppliers and map them to Zendesk organizations by name or external ID.
• •Index open invoices and purchase orders from JD Edwards and make them searchable within Zendesk tickets by supplier, invoice number, and PO number.
• •Authenticate to JD Edwards via AIS Server session token and handle token expiry with automatic re-authentication on the next poll cycle.
• •Enrich Zendesk tickets with ERP context without writing data back to JD Edwards, keeping the support system the source of truth for customer interactions.
• •Respect JD Edwards pagination, rate limits, and IP allowlist requirements while polling on a configurable schedule aligned with your operational rhythm.

Questions

Why is the data flow only from JD Edwards into Zendesk, not both directions?

Zendesk is a customer support platform with no native finance or procurement entities like invoices, accounts, or GL accounts. ml-connector reads supplier, invoice, and PO data from JD Edwards to enrich support tickets with context, but there is no financial data in Zendesk to write back to the ERP. Any changes to suppliers, invoices, or orders must be made in JD Edwards and will be picked up on the next poll cycle.

How does ml-connector handle the JD Edwards AIS Server being on-premises with no public hostname?

The customer provides the full AIS Server URL (including hostname, port, and path) as part of the credential setup, since JD Edwards publishes no shared base address like a SaaS product. The customer must also whitelist ml-connector's egress IP address in their AIS Server allowlist so that polling requests are accepted. ml-connector validates that the service account license remains active on each token refresh.

What happens if a JD Edwards session token expires during a poll cycle?

JD Edwards session tokens last 30 to 60 minutes by default. ml-connector handles token expiry by detecting the HTTP 444 response and immediately requesting a new token with the stored credentials. If the new token request fails, the poll is paused and retried on the next cycle; the enriched data in Zendesk remains until the next successful sync.