Something important to recognize is that each of these business processes optimized and streamlined by digital transformation take either a predictive or reactionary approach to embedded AI. While predictive is more beneficial to a company, it is nearly impossible to always be one step ahead. The best example of this may be the use of embedded AI in cybersecurity trends.
Companies hope to be as predictive as possible when it comes to protecting their devices and data from malware and cyberattacks. The hope is that AI can predict cyber threats before they do any damage to a business; however, there are so many new forms of malicious cyberattacks that it is impossible to predict them all. Therefore, security applications need to be reactionary as well. If a piece of malware does get through a threat intelligence solution, the embedded AI needs to be able to immediately take the appropriate steps to mitigate any damage or potential loss of data.
It would be an ideal world if we had the solutions to all of our business problems before we knew what they were, but that is just not a realistic expectation, so reactionary AI is still necessary. Spaces such as ERP, where the embedded AI provides insights based on historical data, are providing a predictive service based off of the reaction of prior performance, inexact human projections, and unknown or outside catalysts. Same goes for the people analytics provided by AI in HR solutions, along with a plethora of other business software.