Announcing availability of machine learning based Recommendation API inside Dynamics 365 for Operations

We now have the recommendations API integrated and available from within Dynamics 365 for Operations. In the Nov. 1, 2016 update to Dynamics 365 for operations, now you can view the product recommendations inside cloud POS. This is the first time that we bring insights derived from Cortana intelligence suite components using machine learning right within Dynamics 365 for Operations. There is plenty of ground work done underneath the hood to get this seamlessly to you. Pricing You are not charged extra for using this service at this point in time. Microsoft provisions all required resources in Azure based on your estimate. Scenarios

• Product Recommendations can be seen only on the cloud POS
• Product Recommendations can be seen only on the MPOS in online mode
• These can be seen when you view details of a product, they will appear automatically next to 'related products'
• These will also appear on the customer detail page
• These can be seen when you are making transactions on cloud POS or on MPOS when it's in online mode

Note: As soon as a customer is added to transactions, recommendations are personalized to the customers' buying pattern and items in their wish list Components

1. Azure data factory - for sending data (catalog & usage) to the blob storage
2. Azure blob storage - for storing raw and processed data
3. Hive scripts - for cleansing raw data to reduce the bias from 'unidentified' customers buying through POS in the scenario when you want the API to make personalized recommendations
4. All the components for spinning up the recommendation API itself
5. Some other components for making sure all services involved are properly authenticated and have the right credentials

Setting up

1. When your environment is being created through LCS, if you chose you want to use machine learning, recommendations related components are installed in your environment

Once recommendation API related components are available in your environment you need to do following in the relevant legal entity to set it up

1. Go to entity store, find the entity 'Retail sales' and hit refresh. This will pick the demo data (or your data) from your operational DB and move it to Entity store
2. Go to Retail parameters --> Machine learning and turn 'enable recommendations' ON
3. If you will like to see recommendations while making POS transactions, go to screen layout designer and drop recommendation control to transactions screen
4. Run the global configuration job - 1110 (to see recommendations on POS) and channel configuration job 1070 (to reflect changes made to POS screen layout designer)

How does it work?

1. When you refresh the entity store entity, following happens
   • Data in the format required by the API is extracted from AX DB and send to Entity store
   • This data is then used by ADF copy activity and sent to blob storage
   • If there is cleansing required, Hive scripts are run as part of ADF activities and cleansed data is stored in blob storage
   • Data from blob storage is used by the API to train a recommendation model
2. When you turn 'enable recommendations' and run the configuration jobs following happens
   • Model credentials and ID are picked up from the API and stored in AX DB, in the web.config for AOS as well as in the retail server
   • Model credentials and ID are made available to CRT so calls for product recommendations from CPOS and modern POS can be honored

Definitions Cortana intelligence suite Microsoft Cortana intelligence suite is a collection of technologies for information management, big data storage, machine learning and analytics and data visualizations. Cortana assistant on Windows 10, Chabot made with Bot framework, Cognitive services and many other applications utilize one or many technologies marketed under the brand of Cortana intelligence. Microsoft Cognitive services The Microsoft Cognitive Services APIs are a suite of several general-purpose Machine Learning APIs that are made available in Microsoft Azure and can be used for any number of applications. These APIs simplify the whole process by abstracting away the complex machine learning models and the operationalization aspects so that users can focus on real business problems. There are several categories of Machine Learning APIs and the Knowledge category includes the Recommendations API. Recommendation API The Recommendations API  – exposes a general-purpose recommender system capability wrapped in a simple to use REST API. Case studies There are many customers using this API and many partners integrating the API into their e-commerce systems. Here are some published case studies.

1. MEO
2. DutchCrafters
3. JJ Food

Why did we do this? This is a natural evolution both for Dynamics 365 and Cortana intelligence powered machine learning services. We will bring more such services into Dynamics 365 in future. Previous posts Please use this post to understand more about this API in relation to Dynamics 365 for Operations, see video of how a customer uses and what our early prototype looked like. [caption id="attachment_895" align="alignnone" width="1024"] See Product Recommendations on Product details page[/caption] [caption id="attachment_905" align="alignnone" width="1024"] See Product Recommendations on POS transactions[/caption]