Ceres AI - Predict Supply Chain Risk is part of the Boomi Marketplace program. Visit marketplace.boomi.com to browse additional pre-built solutions that you can install directly into your Boomi platform account!

Click here to install this recipe

Ceres Technology offers an AI solution that enables clients to forecast supply chain disruptions and delays, often months in advance. Our AI engine empowers you to identify which raw materials, components, finished goods, and suppliers are at risk of delays or disruptions before you place your purchase orders. This capability allows for better planning and procurement, helping you understand which suppliers are vulnerable, manage inventory more effectively, improve on-time deliveries, enhance supply resilience, and meet Service Level Agreements. Not only do we identify risks, but we also quantify the expected severity of any delays (i.e., how long the delay is likely to last). Our AI incorporates over 25,000 global indicators (as of this writing) including geopolitical information, commodities prices, financial data, global news, and micro- and macroeconomic data. We combine these external sources with clients’ historical purchase order information to forecast potential delays.

The “Predict Supply Chain Risk” AI Agent recipe includes pre-built processes that integrate the Ceres Nostradamus platform (a cloud-based system) with your SAP ERP that do the following:

• Upload new Draft SAP Purchase Orders (POs) to your dedicated Ceres Nostradamus environment and initiate risk prediction processing.

• Receive the Prediction Results back from Ceres, update Draft SAP purchase orders with risk information, and send emails to alert staff.

The CSV file format used to upload new Purchase Orders for prediction processing to the Ceres platform 

Data Element	Description
Purchase Order #	Purchase order number
Supplier	Internal or legal name of the supplier
Product Name	Name of product
Purchase Order Date	The date on which the Purchase order created
Supplier Origin Country	The country of the supplier
Destination State Province	The State of the shipment destination
Price	Price paid for the order
Order Quantity	Quantity of the product ordered
Expected Delivery Date	The date on which the shipment is expected to arrive
Actual Delivery Date	The date on which the shipment actually arrived

Before You Begin - Boomi for SAP Table Service

If you are new to Boomi for SAP, kindly visit this community article & help documentation to get more details. Also consult the documentation for how to configure a Table Service in Boomi for SAP.

You will need to configure a Table Service to read purchase order details from the SAP system.

We have prepared the below sample Table Service (PO_TEST_MOB) using Boomi for SAP which can help you to get the necessary Purchase Order data from your SAP system. (Boomi for SAP also has the capability to share existing Table Services with Customers/Partners).

Configuring EKKO as Master table

• Configure the table service with the EKKO table data(as master table). This provides all the header data of the Purchase Order. 

Joining other tables with EKKO table

• The table EKPO holds all the Items data that particular purchase order contains

• The Table EKET holds purchase order Scheduling data

• The Table LFA1 holds Supplier data

• The Table KNA1 holds Customer data

• The Table MARA holds Material data

• The Table MAKT holds Material Description data

• The Table T001W holds Plant data

Joining all these tables together will provide key data elements related to a purchase order.

Filtering on statuses in EKKO & other tables

• Using Boomi for SAP you can easily filter on fields in your table service configuration, to get a lighter payload, or simply make the data output more comprehensive.

• For example, In this case you could configure a Table service filtered on EKKO table with EBELN (Purchasing document ID), BUKRS (Company Code), LIFNR (Vendo ID), BEDAT (Document Date) etc. This will help the Boomi integration developer when doing the mappings with other tables & also helps to pass only necessary fields to Ceres platform.

• You can also share/copy this(or any) table service (built using Boomi for SAP) with another SAP system using SAP Transport (hyper-automation)

Process: Read SAP Purchase Orders & Send to Ceres Platform

This process picks up purchase orders, transforms them to the CSV format expected by Ceres, uploads the data file, and initiates the prediction processing.

Connector Step - Read Purchase Orders Data from SAP System

Boomi for SAP Connection Details

• URL - http://mysapserver.com

• SAP client - 800

• User Name - Your SAP system User Name

• Password - Your SAP system Password

• Connection timeout - 30000 (for 30 secs)

• Socket timeout - 30000 (for 30 secs)

Boomi for SAP Operation Details

• Connector Action - Get

• Object - PO_TEST_MOB

• Response Profile - Boomi for SAP connector PO_TEST_MOB QUERY Response

• Tracking Direction - Output Documents

• Max Returned Rows - 10

• Batch Size - 1

In the above example we have set the Max Returned Rows to a value of 10, to be used for testing purposes. This will need to be adjusted in order to load an appropriate number of historical purchase orders to effectively train the AI model.

Data Process Step

Set to Combine Documents from the JSON response profile ‘Boomi aXis for SAP PO_TEST_MOB QUERY Response’ returned from the ‘PO_TEST_MOB’ table service to get all SAP Purchase orders under the same JSON structure.

Map Step

Map SAP Purchase Order Profile (JSON) to the Ceres compatible Purchase Order T able Profile (Flat File) as CSV to send it further to Ceres Nostradamus environment.

Set Properties Step

Set Dynamic Document Properties to define CSV File name.

Sub-Process: (Sub) Encode as multipart/form-data

This specific Sub-Process is responsible for creating the received SAP Purchase Orders CSV file request as multipart/form-data MIME type that is needed to send it to Ceres Nostradamus.

HTTP Client Connector Step

• Action - Send

Connection

• URL - https://api.dashboard.cerestech.co/v3/nostradamus/prediction

• Authentication - None

Operation

• Connector Action - Send

• Select Request Profile Type - None

• Select Response Profile Type - None

• Content Type - multipart/form-data; boundary=testboundary

• HTTP Method - Post

• Return HTTP Responses - Selected

• Request Headers

  • Name - ceresapikey

  • Value - Put the API Key received from Ceres Support

If successful, we should expect to receive a JSON response to indicate the prediction processing has been successfully started.

Set Properties Step

It is good practice to ‘SET MIME Property - MIME Document to FALSE’ after multipart/form-data processing to avoid potential issues whereby subsequent HTTP processing still expects MIME formatted documents.

Process: Ceres Prediction Result Listener

This web service listener process is called from the Ceres platform as part of a callback mechanism. Once prediction processing is complete, predictions are sent back to Boomi so that they can be acted upon.

Start step

• Type - Web Services Server Connection

• Action - Listen

• Operation - EXECUTE PredictionResult

  • Connector Action - Listen

  • Simple URL Path - /ws/simple/executePredictionResult

  • SOAP URL Path - /ws/soap- Unavailable

  • Operation Type - Execute

  • Object - predictionResult

  • Expected Input Type - Single JSON Object

  • Request Profile - EXECUTE PredictionResult Request

  • Response Output Type - Single JSON Object

  • Response Profile - EXECUTE PredictionResult Request

Map Step

Map Prediction Results (JSON) received from the Ceres to Flat File Profile so that we can use these mapped fields further to update into SAP & also send an email to end-users regarding prediction results about his/her purchase orders created in SAP initially.

Data Process Step

Set to Split Documents from the Flat profile ‘Prediction Result Flat File Profile’ to separate each purchase order individually, So that it is easier to update each of these purchase order prediction status one by one in SAP .

Branch 1 Step

This specific branch responsible to update prediction results back to SAP against each of the purchase orders which may face delay in delivery.

Message Step

Preparing SAP compatible JSON profile dynamically to update each purchase order (which may get delay in delivery as per prediction results) in SAP .

Boomi for SAP Connector step

This time we are using Function Module Service from Boomi for SAP to update these purchase orders delay status in SAP.

We are using standard SAP FM: BAPI_PO_CHANGE to update these purchase orders with delay status.

Boomi for SAP Function Module Service set-up: Documentation for how to configure Function Module Service in Boomi for SAP

Branch 2 Step

This branch is responsible to send entire Prediction Results to the users as an email.

Message 1 Step

Preparing email format to send in tabular format.

Message 2 Step

Preparing email body.

Set Properties Step

Setting Mail IMAP body content type to “text/html” so that it can prepare the mail in tabular format using HTML tags used in message steps previously.

Mail (IMAP) Connector Step

This particular connector is used to send an email to the users.

If you are new to Mail(IMAP) Connector then follow these steps to set it up: How to Use the Mail (IMAP) Connector

Summary

• Proactive Delay Management: Integrating Ceres with SAP or other ERP systems uses advanced forecasting and AI analytics to predict and address purchase order delays before they impact operations.

• Informed Decision-Making: The system provides valuable insights that enable businesses to take proactive measures, improving decision-making and minimizing risks.

• Optimized Supply Chain Performance: By anticipating disruptions, this integration ensures smoother operations, boosting efficiency and enhancing service levels.