I am often asked the question in the title – what is a smart (enough) system? Here’s the list we use when we talk about it:
While one can make systems of all kinds “smarter” we are talking about making operational, transactional, high-volume, typically customer facing applications smart enough to be useful.
As a result we talk mostly about real-time or near real-time. Rapid batch systems, even slow batch systems, can be smart enough but the trend is inexorably towards real-time.
- Rapidly evolving – agile
No matter how hard you try you cannot make a system smart enough once – it must be an ongoing process. The world changes, business changes, regulations change and competitors change. To be smart enough your system must change with it – it must be agile.
Not in an AI sense but in the sense that it must use the data you have accumulated (about customers, about products, about what works and what does not) to take more appropriate, more profitable actions. It cannot be passive and regard data as something to simply be regurgitated when asked – it must use it.
Both in the sense of taking a holistic view of customers and in the sense of being about how customers are treated. Not all smart enough systems are customer facing – some face suppliers or employees – but it’s a good general principle.
- Extended-Enterprise Ready
No modern enterprise is an island and a system cannot be said to be smart enough unless it reflects this and supports the kind of extended enterprise required these days. It must be possible to treat customers correctly even when they are interacting through a third-party and it must be possible to inject the “smarts” into processes run b y a mix of companies.
- Demonstrably Compliant
Last but by no means least, a smart enough system is compliant and can be easily demonstrated to be so. With more and more regulation and faster response times, a system that is too opaque or hard to control is just not smart enough.
That’s our list. What else would you suggest?