Two articles I saw recently (Is SOA Enabling Intelligent Agents? and Three Keys to Enabling Agile Business Services) made me think about decision services in the context of agility and of so-called “intelligent agents”.
Clearly SOA, web 2.0 and network-centric platforms make “intelligent” and agile agents or services viable but they are enablers – necessary but not sufficient. To build intelligent agents – and I believe that agility and intelligence go hand in hand as intelligent agents must be able to respond rapidly to changing conditions – one needs to thing about business decisions.
Intelligent agents are designed to make choices and take actions on your behalf and do so “intelligently”. They must therefore make decisions as to what is interesting, relevant, worrisome or whatever. They requires decision services or decision agents that can answer questions of relevance or importance.
A decision service or decision agent is focused on a single business decision and has a couple of key characteristics:
- Use of business rules to define the logic of the decision declaratively
- Business user ownership of these rules (or of some of them at least) so that the business can change the logic of the decision whenever it makes sense to do so
- Use of predictive analytic models to turn uncertainty about the future into probabilities that can be acted upon.
- Constant testing and improvement of the decision making approach to make sure that some other approach will not do better – often called adaptive control or champion/challenger or A/B testing
- Optimization of decisions about resources
Decision management and the production of these kinds of decision services or agents is doable now and builds on the power of SOA, web 2.0 and network platforms. They may not be “intelligent” but they are smart enough to be useful.