The principle of yin and yang is that all things exist as inseparable and contradictory opposites. The two opposites of yin and yang attract and complement each other. Neither is superior. While dating back hundreds of years, this ancient concept is still relevant particularly when applied to the interrelationships between modern technologies.

Clients usually approach my team with asks that fall under two broad themes. The first theme relates to their data platform, and the conversation often begins with, “I think I need a data lake.” The second theme relates to artificial intelligence (AI) and often starts with, “How can I find the value in my data,” or more specifically, “How can I use data to reduce production downtimes or drive sales of my new product?”

These two themes signal to me that most clients understand the importance of data and AI and the value they can bring to their business. However, the link between these two themes is the ability to scale AI, which is often restricted due to legacy platforms and siloed processes. In fact, according to Avanade’s AI Maturity research, conducted by Vanson Bourne, 95% of respondents see AI technology and processes as the most important elements needed to scale AI throughout the organization.

I’ve moved my data to the cloud, now what?
Moving data from on-premise legacy systems to the cloud is often a “lift and shift” exercise that delivers the first return on investment. Without a modern data platform that allows teams to derive rich analytical insights, organizations are unable to efficiently link data sets, quickly gain insights, design high value use cases, and use the power of AI at scale to deliver enriched outcomes for the business and customers.

If an organization wants to truly realize the full value from the cloud, it takes a modern data platform to help companies enable AI proof of concepts and then scale them to an enterprise-ready state. Once a foundational data platform is in place, structured and unstructured data is available and can be linked and tool sets can be put in place to efficiently manage AI workloads.

Can I see if there’s value in my data before building a modern data platform?
Clients are often surprised at how quickly they can find value in their data. For clients looking to quickly prove value, we start by identifying the most productive and efficient use cases to solve their problems. Then, we take time to understand what data is required to address the desired objectives.

From there, we run four- to six-week AI/machine learning experiments using relevant data extracts (including external data sources if appropriate) to prove the value in the data in driving the business outcome selected. Once value from the data is proven, we can now industrialize it and deploy it at scale to deliver return on investment, cost savings and/or efficiencies, all while leveraging the power of Azure Cloud Native services and a modern data platform.

According to our research, around a third (32%) of respondents believe it will be difficult to achieve their business objectives without an AI strategy in place. In other words, we should all be diligent in ensuring that data and AI supports modern business goals, which are increasingly digital, and data driven.

AI meets data platform to deliver AI@scale and increased business value
Successful companies focus on building both an AI strategy and a reliable data modernization platform in parallel to driving AI use cases because they realize this will be the catalyst to scale AI across the business and deliver on their strategic goals. For example, a leading boiler manufacturer and smart heating provider wanted to forecast boiler failure service requests to drive improved customer satisfaction. The client wanted to reduce the three-day average response time for a boiler failure request to a same-day response time.

We helped our client by first conducting a three-week Insight Discovery workshop to prove the value in their data and its applicability to predict boiler failure with a high level of accuracy. To accomplish this, we used a combination of hypothesis and new and external sources like weather data to build a blended forecasting approach with 450-plus machine learning models, which predicted boiler failure with an increased level of granularity across regions, fault types, weather fluctuations and days of the week.

The new approach provided the client with a 98% forecasting accuracy and a 60% reduction in unplanned service visits during the summer months. This quickly proved value for the client and led to them implementing a large-scale IoT initiative to improve service models. A digital twin concept was also implemented – a digital replica of physical hardware used. This larger initiative required a data platform and increased value for the business by improving customer satisfaction and driving significant annual savings for the company.

With great AI comes great responsibility
Many clients I work with recognize the risks and the growing importance of digital ethics. In our research on AI Maturity, we found that 95% of business and IT decision-makers think there could be negative consequences from not taking digital ethics seriously.

As companies scale AI, organizations face greater responsibility to address multiple challenges from transparency and privacy to unintentional biases within the data. Avanade has a digital ethics framework, including an AI-specific framework that helps organizations embed digital and AI ethics into their processes, and extend ethics considerations into governance structures, metrics and audits.

Different asks, similar desired outcomes
While not opposites, driving value from AI use cases and building a modern data platform are mutually reinforcing, complementary strategies that overlap and support each other – I like to think of it as AI being the yin to a modern data platform’s yang. One without the other creates a value gap that hinders an organization’s ability to seize and scale the greatest value and return on investment from its AI initiatives.

As an engineer, I know that AI is no longer just a technology in a lab. Success in AI starts with a strategy tied to business goals and high-value use cases that drive differentiated customer and employee experiences, transform products, and automate and optimize operational processes at all levels within the organization.

Most importantly, AI and its ethical use is the responsibility of not just organizations but of each individual. The real risk lies in not having an ethical framework that supports the use of AI now and in the future.