Infusing AI into applications

Applications that run in CICS TS can make more timely and better decisions, and achieve improved business outcomes, by capitalizing on AI within their transactions.

What is AI and why use it on IBM zSystems?

Artificial intelligence (AI) is broadly used in the technology world to describe solutions that can learn on their own. Machine learning (ML) is a subset of AI and encompasses algorithms that make predictions by applying statistical methodologies to identify patterns in past behavior. Deep learning (DL), which is a subset of machine learning, uses neural networks that can, when exposed to different situations or patterns of data, learn on their own. Deep learning refers to a neural network that consists of more than three layers, including the input and the output. You can learn about typical use cases with AI on IBM zSystems at Journey to AI on IBM Z and LinuxONE.

IBM zSystems, and the IBM Integrated Accelerator for AI incorporated in IBM z16, can optimize the processing of machine learning and deep learning algorithms. You can take advantage of these capabilities when using suitable AI models with any supported release of CICS TS.

Why infuse AI into CICS applications?

Infusing AI is about being able to apply AI across your enterprise, drawing on predictions, automation, and optimization to improve your business decisions and outcomes. It's also about making AI part of your day-to-day operations.

By infusing AI models into applications running in CICS TS, you enable real-time decision making within the transactions, significantly reduce latency over making calls off-platform, and avoid the need for the data to leave the platform. The data that provides input to AI models is often relevant only at the time the transaction is being processed: should this customer be approved for a loan at this time? Does the customer's current circumstances make them eligible for a better insurance rate? Can this insurance claim be fraudulent?

How to infuse AI into CICS applications?

There are a variety of methods for infusing AI in your CICS applications. The choice of which option works best for your use case depends on a number of considerations, including the selection of AI model and where it is to be deployed, the response time required by your transaction, and the tools and products that are already in use within your enterprise.

Some of the most common methods for infusing AI into CICS applications are:

Using IBM Watson Machine Learning on z/OS (WMLz): You can make an EXEC CICS LINK call from the CICS application to invoke an AI model deployed to the WMLz scoring engine running in CICS, or via a REST interface to invoke a model in WMLz in a separate address space.
Using IBM Operational Decision Manager (ODM) with WMLz: You can invoke an enhanced ODM rule from the application to reference a model deployed to WMLz and use the prediction from the model in the rule.
Using a community-available AI framework, such as IBM Snap Machine Learning (Snap ML), TensorFlow, or PyTorch: You can make a REST call from the application to an AI model deployed to the AI framework that is hosted either in an IBM z/OS Container Extension (zCX) within the z/OS environment, or in Linux® on IBM Z.

For more information about each option and how to choose among them, see the Planning AI infusion topic in the Infusing AI into applications on IBM Z documentation.