Oracle Fusion Cloud ERP and Microsoft Teams integration

What moves between them

The main flow is from Oracle Fusion Cloud ERP into Microsoft Teams. ml-connector polls Fusion every 5 to 15 minutes for new and updated invoices, receivables invoices, and payments. When it finds a change, it extracts the supplier name, invoice number, amount, and due date, and posts a formatted message to a designated Teams channel. The finance team reviews the notification and takes action in Fusion or the approval workflow; Teams is read-only, so ml-connector does not write financial records back into Teams.

How ml-connector handles it

ml-connector stores both Fusion and Teams credentials encrypted and refreshes Fusion's OAuth 2.0 bearer token when responses return 401. It polls the Fusion REST API using OData filters such as ?q=LastUpdateDate>={timestamp} to retrieve only changed records since the last poll, reducing API load and response size. For each invoice or payment found, ml-connector formats the transaction details (supplier, amount, date) and posts the message to the Teams channel via Graph's /sendActivityNotification or /messages endpoint. Because Fusion Cloud has no outbound webhooks, polling on a predictable schedule is the standard integration pattern. ml-connector handles the Microsoft Entra ID OAuth token refresh and respects Graph's subscription renewal requirement (every 3 days for most change notification types). If a poll or post fails, ml-connector retries with exponential backoff and logs every record attempt so missed notifications can be replayed.

A real-world example

A mid-sized professional services firm runs Oracle Fusion Cloud ERP for finance and procurement across three locations. The accounts payable team, spread across two time zones, uses Microsoft Teams for daily collaboration. Previously, the team relied on email forwarding and manual checks of Fusion's invoice queue; high-priority invoices from key vendors sometimes sat unnoticed, and approval workflows were delayed. With Oracle Fusion Cloud ERP and Teams connected, new supplier invoices post automatically to the team's Teams channel within minutes of arrival in Fusion. The channel message includes the supplier name, amount, and due date, so the AP manager can triage invoices and route them for approval without logging into Fusion. The integration also notifies the team when a payment is due soon, cutting the time spent hunting for deadlines.

What you can do

• •Post new supplier invoices from Oracle Fusion Cloud ERP to a Teams channel within minutes of arrival, including supplier, amount, and due date.
• •Notify the finance team of upcoming payment due dates so critical payments do not miss deadlines.
• •Authenticate Fusion with its tenant-specific pod URL and OAuth 2.0 credentials, and Teams with Microsoft Entra ID OAuth 2.0.
• •Poll Fusion on a 5 to 15 minute schedule filtering by LastUpdateDate so only changed records are fetched.
• •Handle both Teams webhook subscriptions and Graph API message posting, with automatic token refresh and subscription renewal.

Questions

Does ml-connector write invoices or payments back into Oracle Fusion Cloud ERP from Teams?

No. The integration is one-way from Fusion into Teams. ml-connector reads invoices, payments, and receivables data from Fusion and posts notifications to Teams channels. Teams is a messaging platform, not a financial data source, so all writing remains in Fusion and Teams only displays information.

How does the integration handle Oracle Fusion Cloud ERP's lack of outbound webhooks?

Oracle Fusion Cloud ERP has no direct webhook capability without Oracle Integration Cloud middleware. ml-connector instead polls the Fusion REST API every 5 to 15 minutes using OData query parameters to filter by LastUpdateDate, retrieving only records that changed since the last poll. This standard pattern keeps load low while detecting new invoices and payments promptly.

What happens if a Teams notification fails to post or a Fusion poll times out?

ml-connector retries failed polls and posts with exponential backoff and logs every record so the event is not lost. Each attempt is tracked in the audit trail and can be replayed if a poll or post fails, ensuring no invoice notifications are silently dropped.