BLOG: Did IBM Put Artificial Intelligence in a Box?

Brad Miller
Practice Director, Information & Analytics

There seems to be an undertone across organizations to find the “silver bullet” for information insights; a quest for a technological solution that will bring meaningful and timely insights to consumers or thru applications with little to no effort. Artificial Intelligence (AI) alludes to this vision and we have seen significant evolution in the last few years toward simplification of training, running, infusing, and monitoring AI models. Often while doing so, organizations uncovered new challenges such as trusted data, quality of source objects, and reuse of past algorithms and insights.

Although we have not achieved the simplicity / democratization of artificial intelligence to the degree we have seen with business intelligence and dashboarding tools, the evolution of the AI ecosystem is at a precipice, empowering more business users than ever before. One example is where IBM is embedding AI processes within the information ecosystem, enabling auto-discovery of assets and increasing reuse of both data and complex analytics assets. Data virtualization technology allows speed of analysis without replicating data, creating a single pane of glass and abstracting complex relationships between data objects. Data management tools identify, classify, mask, and cleanse aberrations, creating documented and trusted sources from basic reporting to AI application integration. In addition, IBM Cloud Pak for Data integrates a broad spectrum of open source technologies, allowing organizations to leverage and harden concepts and designs that otherwise would not meet security, lineage, and other enterprise production requirements.

The notion of a silver bullet may be more of a dream than reality but the new, end-to-end information ecosystem solutions within the IBM Cloud Pak for Data allow organizations to simplify complex business insights like never before. IBM Cloud Pak for Data offers organizations flexibility of deployment, even delivered in a fit-for-purpose AI in a box form factor called the IBM Cloud Pak for Data System. So, yes, IBM has put AI in a box, but for all the right reasons.

Background: IBM Cloud Pak for Data – Ecosystem Enabling Artificial Intelligence

In October 2018, IBM made a significant acquisition, Red Hat. One of the key motivating factors behind the purchase was Red Hat OpenShift, an application modernization and orchestration solution that allows organizations to build on a unified architecture whether the solution needs to run on premise, in the cloud, or a mix thereof. By leveraging containers and microservices, the architecture allows for much more efficient use of infrastructure and optimization of processing across the ecosystem. Common logging, administration, and other services can be abstracted and shared across containers, giving more ease of interoperability than with monolithic solutions. The IBM Cloud Pak for Data builds upon this revolutionary foundation and brings a broad universe of functional components to an organization. The beauty of the design is that organizations only need to spin up and leverage the functional components that are relevant to the use case, project, or program needs.

The IBM Cloud Pak for Data supports organizations on their information and AI technology maturity journey, empowering the refinement and management of information across a broad spectrum of solutions. IBM continues to add functionality iteratively, but as you can see below, has taken proven technologies and brought them under the umbrella of IBM Cloud Pak for Data. Organizations that purchase Cloud Pak for Data are licensed to use any of the core components, which cover everything from data collection and repositories (such as Db2 Warehouse) to virtualization, DataOps, governance, and many forms of dashboarding, analytics, and AI such as IBM Watson. Many of the products consolidated into Cloud Pak for Data have been successful standalone products, but have now evolved to bring their solution strengths together into one homogeneous platform.

The IBM Cloud Pak for Data System streamlines complexities, interlacing compute, storage, networking, and software to deliver an expandable and extensible architecture and shortening the time to get to the insights. Organizations that have implemented the Cloud Pak for Data system have found that their teams are refining data and driving insights with AI technology in a few days and showing a return on investment almost immediately.

Business Success – AI in a Box Use Cases

Mainline has been on the forefront of the Cloud Pak for Data evolution, working with IBM and our customers to implement successful solutions across industries. The Cloud Pak for Data breadth of functionality allows for a diverse set of use cases and challenges to be addressed within one platform. Mainline and the IBM Cloud Pak for Data system have quickly enabled organizations in the industrial, retail, banking, financial services, healthcare, and research verticals to meet their specific data and AI needs.

Insurance – Loss Mitigation

Insurance companies by nature are tasked with assessing and mitigating risk. There are many innovative means that organizations have used to accomplish this but with the advent of AI and machine learning, new opportunities are now available. By leveraging the insureds’ equipment and vehicle information, usage patterns, real-time weather information, and text alerting, Mainline’s client was able to predict severe weather and alert / highlight relevant information to its members. By sharing a simple message highlighting the threat and giving potential safe zones, the organization was able to mitigate and quantify loss while at the same time increasing member satisfaction. Members who did not even use the organization’s vehicle and equipment coverage were still able to opt into the program and gain value by using the information to support their decision making.

The IBM Cloud Pak for Data System allowed this organization to quickly ingest, organize, qualify, model, and infuse insights into forms in which their members saw immediate value. Satisfaction from members who were not only able to save damage to equipment, but also significant time and effort dealing with the damage, allowed the program to return 2x the investment after the first year.

Healthcare – Genomics Research and COVID-19 Research

Healthcare organizations are dynamic, as the equilibrium between expanding costs and delivering optimal service are held in delicate balance. One large regional healthcare and research organization implemented the IBM Cloud Pak for Data solution early in 2020. The initial goal was to modernize and empower over 120 researchers and 1200 consumers with information insights supported by machine learning, including COVID-19 pandemic research, modeling protein structures that could slow or stop the replication of the coronavirus. This organization had been leveraging a legacy Netezza environment and within two weeks, Mainline was able to transition them to the Cloud Pak for Data system with Netezza Performance Server. Once the core data lake and AI processing engine were transitioned, the organization was able to gain additional value by consolidating the researchers’ work areas into a unified collaboration zone where reuse and visibility into the organization’s data objects, attributes, and AI artifacts are now possible. The Cloud Pak for data architecture, including Watson AI, allows for broader data processing including unstructured data and Natural Language Processing (NLP). The combination of these factors has allowed deeper insights in shorter timeframes.

Banking / Financial Services – Regulatory Compliance and Customer Satisfaction

Banking and financial services organizations have been under significant pressure in recent years with ever-changing regulations and market pressure. Dodd Frank and Comprehensive Capital Analysis and Review (CCAR) provisions for larger institutions require predictive asset and risk assessment in addition to transparency, accountability, and other regulatory provisions, which can result in significant cost and effort. Although necessary, these regulations can often be their biggest challenges, slowing or causing distraction from the organization’s core business growth and cost management functions, such as fraud detection and other consumer-facing services. A regional bank working with Mainline has implemented the Cloud Pak for Data solution in order to accomplish the goal of meeting regulatory compliance while at the same time expanding consumer-facing initiatives, including AI systems-based fraud detection and mitigation as well as enhanced consumer offers to increase customer satisfaction. The Cloud Pak for Data platform also allowed this organization to modernize their entire information and analytics ecosystem, increasing quality in their data center and accelerating data processing capabilities to near-real time. The Cloud Pak for Data solution integrated seamlessly with open source technologies and empowered broader use through the enterprise hardened and governed capabilities across the platform. The cataloging, lineage, and governance capabilities have given both internal staff and regulatory auditors Google-like functionality to find and understand the organizations’ data assets and artifacts along with any transformational journey that data took to get to the end consumer or application. This organization has already recovered the investment in the Cloud Pak for Data solution by deprecating legacy systems and now, with real-time analytics, fraud mitigation, and enhanced customer satisfaction, the organization has reaped qualitative benefits as well.

Apparel / Manufacturing – Customer 360 / Product Targeting

The Customer 360 use case is a common target for modeling and AI workloads. Adding in real-time, next-best product offers and tying it all back to inventory and even production brings a breadth and complexity to the AI models and processing that is quite significant and often costly. One of our Mainline customers had a directive to move to the cloud and thus picked up and moved both applications and analytics workloads to Microsoft Azure. Their data scientists were tasked with building a near-real-time, enhanced Customer 360 model and infusing the data into the organization’s consumer-facing applications as well as marketing and operations systems. Although significant tuning and consulting hours were invested refining the models, there was no clear way to get to the targeted outcomes without expanding the processing capabilities on Microsoft Azure to the point where the cost was too high. Mainline created a solution with IBM Cloud Pak for Data that leveraged the code base deployed on Azure and processed it through a high-performance, on-premise architecture. This hybrid architecture allowed for an optimal balance of cloud and on-premise processing and the cost of the investment was returned in three months’ time.

More Information:

For more information on artificial intelligence and machine learning solutions, contact your Mainline Account Executive directly or click here to contact us with any questions.

You May Also Be Interested In:

BLOG: What Happened to Netezza

Webinar Replay: Cloud Pak for Data Turning Data into Insights (54:32)

Video: Cloud Pak for Data – Make your Data Ready for AI & Cloud (2:09)