We did portals, then came content and collaboration, then came the big platform SAP NetWeaver, then came apps on NetWeaver - xApps, Composites, Analytical apps, etc. - but what is next?
Lets start with the obvious with NetWeaver - People, Processes and Information.
People: The portal, WebDynPro, enhanced UI, Collaboration, Chat, email, personalization, etc. Once deployed these sets of technologies are all about how people use applications. Therefore at some level these apps have a repository of user patterns.
Processes: The technologies in NetWeaver allow abstraction of existing business processes from underlying systems and repurposing them to create newer processes. At some level there is a layer of meta data that holds information on what objects exist, which of these are connected, who is using them and to a certain level, the strength or weight that these processes hold. All of these are patterns.
Information: This is the easy piece: Unstructured Content and Structured Content.
The basis for most AI tools in the market make models based on historical data. They make rules/models based on historical data and make logical predictions. And this becomes a cyclical process, as the data is qualified the models become stronger.
My prediction is that as NetWeaver grows we are potentially sitting on a wealth of data with tons of Patterns. And these patterns are at all levels - User, Process, and Data. And the most interesting part with NetWeaver is that these patterns can be connected. For example starting with the user there are patterns with how he looks at and uses business processes and data.
If we move a level up and say that there was an AI tool embedded with NetWeaver, then this tool would constantly look at the interaction of business processes and objects, users rights with these processes, how these interact, what kind of reports users are looking at, and possibly their relations to business processes. We could also look at the relation from data and processes to unstructured content - all of this based on aggregation of patterns at all stages within NetWeaver.
An idea does not make sense if we cannot provide some practical examples on how this new addition can be made use of in real life.
The idea at the very least takes "suggestion" to the next level. Automated suggestion is a good start, look at the way Gmail works. For the business user based on his rights, user patterns, roles, data in emails, etc., automated suggestions can be made to enhance his use of business applications. Another obvious area is fraud prevention; AI already plays a part here with Credit Card Processing, but in our case having this functionality to monitor business process rights will have tremendous impact on audit, SOA compliance, and a whole bunch of other areas like IT security.
In the manufacturing sector AI plays a role by automated monitoring of MES and DCS systems at the plant floor, but combining them with how resolutions can be automated or suggested has huge impact on productivity. Some of the other areas I can think about are Vendor Performance, Contract Compliance, Compliance in general, Military deployments (AI+NW+RFID+SCM), pharma and the list goes on... leaving the rest to our imagination.
The underlying idea is to be able to send basic business transactions, alerts , etc., thru these models (collected from NW based on historical patterns) from this AI Engine. This would give us a list of relevant possibilities that can be sorted by relevance. The advantage of having an AI Engine is that these possibilities can be sorted by the interactions' relevance, and without a human interface. Rules can be set on top to either kick off an automated command or to alert a user. For those of you who understand Unifiers, there used to be a way to co-relate objects and assign weights to them. The problem here was that these created endless possibilities and the scenarios which made most sense were to be created manually. With the AI engine it would be possible to do this automatically based on patterns!
