AI platforms are not your typical IT systems, built on a set of rules and explicit logic that provides programmed functionality but no new insights. Rather, they are “Systems of Intuition” that learn from past and current data in order to build their own reasoning, providing value beyond what is normally expected. As such, starting the journey to build your own AI platform is different from a typical “digital transformation” undertaking. The first step to a successful AI platform implementation is the identification and selection of the business use case which the AI platform will help solve. The selection of the business use case will affect most if not all of the subsequent decisions that are made as the design and development of the platform progresses. Selecting the right use case is fundamental to ensuring the anticipated ROI of the platform is realized.
The best source of business use cases is as the name suggests – “the business.” The business is integral in driving the process of identifying real opportunities where AI could solve problems that traditional IT systems cannot. For example, providing human-like predictions, insights, and recommendations, augmenting an existing IT system with AI capabilities, or alleviating a particularly worrisome business “pain point.” It is advised that business leads the use case conversation with the participation of the technology team and other stakeholders.
Once the proper team has been assembled, there are two ways to get to the right business use cases:
Follow the Data
AI models and platforms require data. The more data, the better. The first method of starting the AI journey is to use the “Follow the Data” path. The team would start by noting all of the likely business use cases. Then, they would assess the enterprise data availability and readiness based on the business use cases to see where there are shared data clusters. Those use cases with the highest data clustering are likely the best use cases for an AI implementation. At the end of that exercise, the team would be left with 4 to 8 very strong business use cases with the required enterprise data identified. The path to AI implementation success then becomes clearer.
Follow the Business Impact
The second path to AI implementation success is to “Follow the Business Impact.” Again, the team would begin by noting all of the likely business use cases. This time, instead of using the enterprise data view as the lens, the team would determine which of the use cases could provide the largest business impact in the following ways:
This method also allows the team to zero in on the best 4 to 8 business uses cases by scoring the various business impact categories based on their own priorities. The final score gives the business a tangible quantitative view into the strongest (and weakest) AI use cases. Of course, to further refine the priority and strength of the selected use cases by next applying the “Follow the Data” filter. This may then provide a final list of 2 to 4 very strong use cases with the highest business impact and clear data availability.
These methods may be new to some organizations. Typical IT projects start from a different need, for example providing business reports to senior stakeholders or regulators. Some of those IT initiatives may also be generated from technology issues, such as the reduction of technical debt. In such cases, implementing AI is not normally a driver for those initiatives. No matter the business or industry, AI is here, and it will find a home at many levels of the enterprise. To begin your own AI journey, it helps to start with the right mindset. A different mindset. One that starts with the business and data rather than technology.