I just saw Charles Nicholls' article "Building Intelligent Business Processes into SOA". This was very timely as not only have I recently blogged about SeeWhy and BI 2.0 before I also saw Charles last week (we used to work together). I agree with a lot of what Charles has to say, and what he is doing at SeeWhy, but I did feel like I wanted to make a couple of points in response to his article. Charles has a great definition of an intelligent process:
Intelligent processes are created thorough the automation of repeatable, operational decisions by embedding business intelligence (BI) into those processes.
Intelligent processes are relevant, personalized and responsive.
Now I might quibble with BI (preferring analytics as reports don't work here, only analytic models) and I might also say that an intelligent process should be one that is compliant with regulations and embodies the best expertise your organization has (both uses of rules) but it's a nice definition anyway. "relevant, personalized and responsive" would be covered by my phrases "precise" and "agile". I do think that intelligent processes, or at least the decisions within them, need to be consistent (across channel for instance) unless you design them not to be. Then Charles goes on to take a dig at rules:
Predetermined business rules and logic governing analytic processes don’t adapt automatically, either, as the business changes or to apply more tailored logic to each process instance. The result: rigid, policy-based approaches that don’t treat different processes’ instances in a relevant, personalized or responsive way.
and he's quite right "rigid, policy-based approaches" do not work. However, not all processes can adapt automatically - if you had a lot of loans going in to default and so your system changed the approval criteria for loans you could find yourself in very hot water. Hence the need for rules (to handle the regulations and policies you want to enforce) and analytics (to handle more adaptive/information-centric pieces). Indeed I wrote about analytically-driven processes and transaction-centric decisioning. I also wrote about using business rules to bring business intelligence into processes before.
I think overall that my attitude to decision automation and Charles' approach to BI 2.0 are very compatiable. I think a little more about managing decisions as distinct assets and about compliance/regulations than he does and he is more focused on how to turn events into insight not just data. When all is said and done though I think there is a role for SeeWhy's kind of decision-making as well as localized rules-based and analytic-based decision-making in Complex Event Processing (CEP) or BAM, but this is not the same as taking control of your operational decisions, treating those decisions as a corporate asset and automating them.
Technorati Tags: analytics, BI, business intelligence, business process management, business rules, compliance, data, decision automation, event driven architecture, event processing, personalization, predictive analytics, SeeWhy
Subscribe to this blog by RSS
Hi James
Thanks for the blog - you may have misinterpreted my intent, so perhaps this is a good opportunity to clarify what I was trying to say, even if it didn't come across!
The approach of building BI into processes is highly complimentary (and relies on)both BPM and Business Rules, and it is where compliance with regulation and policy are usually enforced. So I wansn't really addressing the compliance point, but aiming to get people thinking about not just automating 'dumb' processes. After all if you're going to the effort of automating the process, let's try and make the process run better, not just faster.
I wasn't having a dig at rules engines, but was explaining how, in order to make rules personalized, the facts on which the rules make their decisions, in many but not all cases, need to by dynamic. We see great potential to use real time BI alongside Rules engines & BPM to asert facts, and as a result to make the decisions which are being executed by the rules or BPM engine smarter.
This doesn't mean that you don't need fixed rules, of course you do, but that real time calculations can add great value. Typically these types of calculations cannot be done by rules engines or BPM products - they're just not designed to do it. This is the 'information centric' part of the decision that you describe. Fair Isaac, of all companies, knows this well! I think we agree.
So to take your loan approval example, the dynamic element might be the price of the loan offered based on the individual value of the customer, while the regulation elements are clearly not going to adapt.
"BI" in my parlance includes analytics as a subset. Since nobody in the industry can agree on what an analytic is anyway, it seems a mute point....but I agree, it's not a report!