The workflow starts when new orders are created in Shopify, using a scheduled trigger to check for new records every 3 minutes.

Filter incoming orders to process only those with a valid payment status.

Check whether the order contains a valid purchase order (PO) number. If a PO number exists, proceed to merge and process the B2B order. If not, create a draft order and send a notification.

Merge the filtered paid order with PO number data using an inner join to combine both datasets for downstream processing.

Look up the business partner in SAP Business One using the customer's email address from the Shopify order.

Retrieve existing open invoices for the business partner from SAP Business One to assess credit exposure.

Use OpenAI to perform credit control analysis by evaluating outstanding balances, new order amount, total exposure, and payment terms compliance. Returns a decision: approve, approve with warning, or apply credit hold.

Convert the OpenAI chat completion response into structured JSON for downstream decision-making.

Based on the AI credit validation result, decide whether to create a sales order with a credit hold flag or create a standard sales order.

If the credit check indicates risk, create a sales order in SAP B1 with a credit hold flag and send an alert email to the customer regarding the credit hold.

If the credit check passes, create a standard sales order in SAP B1, update the Shopify order notes with the SAP document number, and send a confirmation email to the customer.

If no valid PO number exists, fetch the business partner by email, create a draft order in SAP B1, and send a notification to Microsoft Teams alerting the team about the invalid PO number.