International Business Machines Corporation (IBM) Earnings Call Transcript & Summary

Welcome, everyone, to our continuing summer series on modernizing your data and AI platform. My name is Russ Milano, and I'm a global executive in IBM's data science organization. Today, we're going to focus on the impact that a modern data and AI platform can have on your data science organization and their productivity and how you can actually lower your current costs while accelerating your data science initiatives. Along the way, I'll share some stories from real clients that have already made this move. In the second half of the webinar, I'll show you how you can save up to 50% on what you are already spending on your data science software today at the same time you're modernizing and accelerating your journey to machine learning and AI. [Operator Instructions] And connect with me in LinkedIn for updates on all things data science. Now to frame our conversation, we need to start with being honest, honest about our current software, our tools and our infrastructure that we have accumulated as companies over the past few decades. When I was much younger, I had the most incredible entertainment system. I purchased the best receiver in the market. I had a large, 42-inch wide-screen TV. I had my stereo system connected: CD player, a turntable with a diamond-tip needle, TiVo for DVR, a DVD, a Blu-ray player and 2 gaming consoles. My friends love my system. They were envious. Now they didn't know that it took me the better part of a weekend to connect it all together. And every few months, I would try to upgrade a component, and that caused issues with other components and wiring. But at the time, I thought it was awesome. It was the best of breed, and I thought I had the best system around. Like I said, I was younger and less wise then. Now today, I have a smart TV. It gives me access to all my content from multiple streaming services. I can access the Internet, and I can add new services in minutes. Overall, it is so much faster and easier to use and less costly than my best-of-breed system. So what does this have to do with data science? When I talk with clients and we discuss their goals for accelerating their data science efforts, we often discover that their information architecture looks a lot like my best-of-breed entertainment system. Various groups and lines of business have purchased or influenced the purchase of a variety of data science and business intelligence tools. Many companies have more than one cloud provider, numerous databases and several storage technologies. Now these may have been best of breed at the time or just the best price, but we have spent several decades now creating information architecture that adhered to the requirements of that time, and that model has introduced a reality that is difficult to scale and will continue to hold us back. Looking at this complexity, it is no wonder that our valuable IT resources are overloaded. How much time do we spend connecting this disparate infrastructure? And how many nights and weekends are spent patching, upgrading and maintaining them? How long does it take your data science and analytics team to connect their tools to the various databases, storage and clouds that contain the data they need? A big question for our conversation today: What's the impact of this infrastructure on your artificial intelligence and machine learning initiatives? How does this inefficient information architecture slow your data scientists and impact the accuracy of their models? We haven't even touched on the extra costs from license fees, maintenance, hardware, the cost of people to maintain the environment and the training costs for those people and the costs of duplicating and migrating data. So what if you could modernize this environment like the smart TV? And what if you could actually save up to 50% in the process? That's what we're going to talk about today, but it starts with redesigning our information architecture strategy. IBM recognizes the challenges our clients are facing. And as a result, we've built a prescriptive approach known as the AI Ladder. This is to help clients overcome the challenges and accelerate their journey to AI. The AI ladder has 4 steps. The first step, make data simple for everyone. So collect the data of every type, regardless of where it lives, enabling flexibility in the face of ever-changing data sources. Second is to organize it and create a business-ready analytics foundation to organize all your data into a trusted, business-ready foundation with governance built in, protection built in and also compliance. Then analyze, analyze your data in smarter ways and benefit from AI models that empower your team to gain new insights and make better and smarter decisions. And then operationalize AI throughout the business across multiple departments and within various processes, drawing on the predictions and automation and optimization models built in the analyze step. All of this is built on Red Head OpenShift, so it can operate on any cloud environment you choose. That could be your internal private cloud, Microsoft, Google, Amazon or IBM, so any cloud or multiple clouds. In the new world, everyone in your company will be able to access data through an enterprise insights platform. This platform democratizes information through an approach that allows users to ensure their data is ready to build AI. This is why IBM has invested so much in Red Hat OpenShift, stress connectivity and interoperability with many cloud vendors and why we have to re-strategize on how we are delivering all of our data and AI offerings. Now let's talk about a client that has started modernizing their data and AI environment. Wunderman Thompson Data is the data science part of Wunderman Thompson, which is a leading digital agency with over 200 offices in 70 countries. They are a pioneer for using data and AI. The Wunderman Thompson Data team of data scientists, analytics practitioners, strategists, marketing technologists and data consultants is still the why of complex data into powerful insights. This empowers the Wunderman Thompson brands with new ideas that strengthen customer relationships and anticipate what customers need today and beyond. Now they had 2 business goals: shorten the time it takes to provide clients with lists of target customers and also improve the accuracy of those lists that they provide to their clients. So they had to innovate to meet the needs and demands of their business and their customers in this challenging environment. Now to do this, their modeling framework needed to run fast, and it needed to run at scale. They need to run thousands of models each year for clients. And sometimes, they're expected to turn around those models in 24 hours. Now one of the big challenges for Wunderman is the amount of data and the number of features in their models. They have 17,000 features to choose from for each person in their database, and their database has 270 million people in the United States alone. Another challenge is that 70% of their time was spent staging, structuring and moving large amounts of data from thousands of data feeds that they receive each day and week. And to add through this, they had to solve this while providing transparency to how the decisions are made, how the lists were built. There's no black box allowed in their machine learning. So Wunderman Thompson Data partnered with IBM to leverage our data and AI platform and our AutoAI capabilities as part of the new foundation for their new information architecture, in essence, their new smart TV. Within a few months, Wunderman Thompson Data and IBM's Data Science Elite team solved a challenge they were trying to solve for 8 years, and that was a challenge of big data. During the recent IBM Think 2020 virtual conference, DBS, the lead data scientist at Wunderman Thompson Data, shared that when they were running on SAS, they had to train their models on thousands of rows of data, and now they can do it with millions of rows. They used to only be able to use hundreds of features, and now they can select from thousands of features. So through the relationship with IBM, Wunderman Thompson Data fully utilizes all of their data signals to produce models that increase the performance over previous models by up to 200% and even more. And instead of spending 70% of their time structuring and organizing data, now they spend the majority of their time doing discovery and building new features. On a recent IBM webcast, Michael Murray, President and Chief Product Officer for Wunderman Thompson Data, summed up their success this way. He said, "We are now just turning the corner into a world where we can go from hypothesis to test up against 270 million people in the U.S. Something that took us months, we can now do in 4 hours." Months to 4 hours. What would that do for your data science initiatives? Now to hear the entire interview with Michael Murray, you can go to ibm.biz/cxowebinar. This is an interview between Michael and IBM's Chief Data Officer for Data and AI, Seth Dobrin. And to hear from Wunderman Thompson Data's lead data scientist, DBS, you can go to the link at the bottom of this page that will take you to the replay from the IBM Think 2020 virtual conference. DBS gives a more in-depth discussion of the challenges and their solution. He describes the challenges they had when they were using SAS and the successes they are now having with IBM Cloud Pak for Data and IBM's AutoAI. So there's 2 key takeaways from the Wunderman Thompson Data successes. First, the IBM integrated data and AI platform called Cloud Pak for Data dramatically accelerates the time frame for data scientists by simplifying and eliminating mundane tasks; and second, IBM's AutoAI is a game changer. A recent Forrester study showed $3.2 million of value is created from the typical enterprise, and those enterprises can realize a 6% improvement in model accuracy, which can mean millions, possibly tens of millions in additional revenue or benefit depending on the application, a 6% improvement in fraud detection to be worth tens of millions to insurance companies and a 6% improvement in predicting sepsis to save thousands or tens of thousands of lives annually in the United States alone. So the benefits are there monetarily. And you've heard about the incredible acceleration in time to value that Wunderman Thompson Data achieved in just a few months. So now the question is how do you pay for this, especially in the tough economic times that we find ourselves in? Consolidating on IBM Cloud Pak for Data can save clients up to 50% and on their annual data science software costs. Let's take an example of a client using SAS that spends $2 million annually. We've helped a lot of clients in this scenario. We help them save money. We help them save time, and we help them eliminate the dependency on SAS. And we do this by allowing them to build and maintain scripts and models in open source, SAS language, or both, on our modern data science platform. They get to choose. And how do we do this? We allow them to import and run their existing SAS scripts and models without changing them, and then they can build new models using our AutoAI capabilities and not relying on SAS. Or if they wanted to continue to build SAS, they could do that. In doing this, they've saved 50% as compared to what they would normally pay SAS annually. Now let's dive a little bit deeper. First, we help clients save money, saving 50% on the total cost of ownership, and they do this because they can run their SAS code without the need for the expensive SAS licenses. They also eliminate the cost that would normally take to rewrite millions of lines of SAS code because they can import it directly into Cloud Pak for Data. We allow them to save time because they can now reduce the time to value by 80% to 90% using AutoAI. They can make data scientists and business analysts a lot more productive. And again, there's no need to spend time manually rewriting the SAS code. And the third benefit is eliminating the dependency on SAS. Some customers complain about the vendor lock-in they have with SAS, and they want to enjoy the flexibility and innovation that comes with open source. So this helps them eliminate the need for SAS licenses and the requirement to train new workers to use SAS as a proprietary coding language. In essence, they can develop in the latest open-source tools they want to use and continue SAS that they'd like. And then for new models, they can use IBM's AutoAI, which generates Python code, so there's no vendor lock-in with our solution. It also allows them to access large pools of talent within the open-source community. So allowing clients to build using their tools of choice is what we found most customers want. They want to be able to use open source. They want to be able to continue to do some work in SAS. They want to use other tools, and they want to do it on a common data science platform. Our approach allows this flexibility and allows clients to accelerate their data science initiatives, and they can use the skills their employees already have. Data scientists and business analysts can be 5x more productive using AutoAI. Existing SAS experts contain -- maintain legacy scripts and models in SAS language, and they do it all on the same platform where their data lives. And it's already governed and collected and organized in a way that they can use it. So a little more detail on how we deliver. We actually allow clients to import and run the existing SAS scripts they have without changing them. This means you can run and maintain legacy SAS code as is. You can eliminate the need for expensive SAS licenses. You immediately have access to all the latest open-source tools in AutoAI. And in our package, we also include some services to install the software and provide up to 3 months of migration assistance. If model changes are required at a future date, then you can edit the SAS code in the JupyterLab's notebook. Or if you wanted to go ahead and rewrite it, you can. You can also build new models using powerful AutoAI that generates the Python code, so there's no black box, no vendor lock-in. And one of the key things is, with our solution, you now get access to the powerful AutoAI capabilities, so you can build new models without relying on SAS. And the AutoAI capabilities will actually produce Python code that your data scientists can inspect. And if they want to improve upon it, they can do that before moving into a production mode. And this eliminates the need for purchasing additional SAS licenses if you have to build new models, which some clients choose to do. To learn more, you can click on the link at the bottom of this slide and try our AutoAI capabilities out for yourself. Another area I want to address is skills. One of the greatest challenges for growing AI is talent. AI skills are rare and therefore in high demand, and there's a shortage of skilled workers that are available to hire. So this makes it even more important that the technology being built and used is more easily accessible for everyone in the business regardless of the skill level. In order to help our clients on the AI journey, IBM offers a number of talent and skills programs to help clients build skills and applied enterprise AI. Some of these programs are our Data Science and AI Elite team, which can help you kick-start your AI project. They're a team of expert consultants in the field of data science and machine learning. We also have our AI learning and certificate program. This allows data scientists and AI practitioners to access AI learning content at no charge, online. And it contains curated content on AI learning, cutting-edge research and links to hands-on courses. We also have expert advice, where you can schedule a free 30-minute consultation with a data and AI expert. And also, we have our DataFirst and Garage method. Wherever you are on your journey, the IBM DataFirst method provides strategy and expertise to accelerate your business outcomes with data and analytics. Earlier, I mentioned a study conducted by Forrester Consulting. Some of the key benefits of moving to IBM Cloud Pak for Data were that it reduced the infrastructure management effort by 65% to 85%, and the data science benefits were between $1.2 million and $3.4 million. And those data science and machine learning benefits are driven by the fact that Cloud Pak for Data helps data scientists improve productivity through faster model development and deployment. Deployment is key. Additionally, due to Cloud Pak for Data's integrated platform, companies avoided the costs associated with legacy analytics tools or otherwise building comparable integrated data management and data science solutions internally, in essence trying to build their own best of breed. One more thing before we wrap up. Now is not the time to invest in a start-up for AI. IBM has been a pioneer in data science and AI, and we weathered the storm for over 100 years. No one else in the marketplace has the position and capabilities that IBM does. Our customers are voting with their dollars, and the analysts agree that we have the #1 offering for data and AI platforms. IBM is a data science leader in multiple recent analyst reports and has won recent awards such as the AIconics for AutoAI capability and iF DESIGN AWARD for our Watson Studio capabilities. So I want to thank you for joining me today and for all the questions you asked in the Q&A. Based on the discussion today, I suspect you're going to want to learn more. So I'll leave you with 3 things. First, connect with me on LinkedIn. I share 1 to 2 topics a week that I'm sure you'll find useful. Second, learn more about Cloud Pak by visiting ibm.biz/cloudpakexperience. Try it for free and even book a free consultation. And third, most importantly, schedule time with your IBM team to learn more and discuss your goals for the rest of 2020 and 2021. Your IBM team is your guide to help you on this journey, and they are trained and ready to help. If you don't know who to contact at IBM, please message me on LinkedIn, and I'll be sure to connect you. So please don't hesitate to contact me or your IBM representatives to learn more and discuss your use cases. Thanks so much for your time today. Stay safe. Stay healthy. Thank you.