The nice folks over at IDC let me take a look at IDC – Worldwide Application Development and Deployment: 2010 Predictions. I have a somewhat narrow focus (Decision Management) so only two are really in my zone. First, #10:
10. Decisioning Products Will Be Unified to Deliver a Decision Management Platform
This is obviously right up my alley and something you might expect me to agree strongly about. And, at some level, I do agree. I agree there is a natural synergy between business rules and analytics – analytic techniques can improve the rules being used and can provide predictions against which rules can be written while business rules make analytics actionable (something that showed up pretty strongly in the research I recently did for BeyeNetwork). I also agree that business rules and CEP (Complex Event Processing) overlap in terms of their use of executable rules to make decisions. More importantly the move to automate decisions is, for many companies, tightly coupled with a move to real-time. This makes thinking about event-driven architectures and event correlation a common parallel activity for organizations using business rules to automate decisions.
Where I am not sure is in the focus on a platform. I agree that we need these products to come together and we need folks to develop an effective decisioning or decision management platform. We already see this with IBM (with WebSphere Business Events, ILOG Rules and SPSS) and Pegasystems/Chordiant, while folks like Progress, Tibco and others are rapidly doing the same. What I don’t believe is that people will BUY a platform for some time. In 2010 and 2011 I think the vast majority of organizations acquiring decision management technology/platforms will do so in the context of a pre-configured or at least template-driven solution: Decision Management allowing claims automation or customer-centric 1:1 marketing or automated loan origination. Most folks I speak to are not yet at the point of realizing that a common decision management platform is something they need – they want the consequences of such a platform in terms of automated decisioning but they are not yet buying the platform.
The other interesting one for me was
6. Configuration-Based Development Becomes an Important New Pattern for Application Development
I would say that any real configuration-based approach to development requires business rules to allow for the “configuration” of logic. If you use code or macros you are just going to add complexity or limit the configurability to simple stuff. At least, without rules/automated decisions, a configuration-based approach is going to be largely constrained to CRUD (Create, Read, Update, Delete) and some workflow. This of course is the problem with many current ERP/CRM systems – they are as dumb as toast: all they do is manage the data and tell people when a decision is required. To be a “smart (enough) system” as Neil and I put it, they need to do more. If your configuration based approach cannot “stray too far from the ER [entity-relationship] model” as IDC puts it then it is going to create new, but still dumb, applications.
Finally, I also checked out the list of development technologies that companies listed as most likely to be used more next year. As an absolute percentage of responders, web development tools, app servers and messaging technologies were top. But if you only included those in each category those who were using the technology already, analytics and data warehousing leaped to the top. Lots of people are not using analytics, but those that are plan to use it a lot more next year. Perhaps those already getting an analytic edge expect to do so more while those who don’t get analytics will keep on not getting it 🙂 Food for thought.
It’s an interesting set of predictions (the other 8 are interesting too, just not in my space) and you can get the report here.
| who based development requies rules |
| or at least without rules/decisions it is constrained to CRUD and some workflow – problem with current ERP/CRM systems |
| “stray too far from ER model” |
| agree on natural synergy between rules and analytics |
| rules/cep overlap but more importanlly move to decision tightly coupled with more to real-time |
| Platform? |
Comments on this entry are closed.
Hi James, I have inferred from your comment: ‘I agree that we need these products to come together’ and from the examples that you give that you endorse the idea that decisioning needs to ‘come together’ with event management (i.e. process) platforms and BI environments. Am I correct in assuming that coming together implies a degree of integration?
If my interpretation is correct then I think that it is contrary to your next point re configurable platforms. When things ‘come together’ in the way that Pega rules come together with the Pega environment, and Ilog comes together with Websphere for instance, you build in tool and design dependencies that create inertia – the enemy of agility.
I 100% agree that the move to configuration based development (in my view this includes model driven development) is the way of the future, and that this must include configurable or model based business rules. For your interest we have just embarked on such a development in the Investment Funds space – since we have just started, I intend to diarize it more or less fortnightly, so we can see first-hand what a model driven platform development that is synchronized with model based decisioning can achieve. You can see it here.
In the approach we are taking, the model for the platform sees the model for the business rules as ‘content’, not as part and parcel of its own design. This implies two distinct modeling environments and in reality has two distinct sets of everything: users (model builders), approaches and development life cycles, model artifacts, executables, inputs and outputs. In fact they only need to ‘come together’ through a communication from the platform to the decision executable and back again – and these can realistically be different executables running on different architectures owned by different players on different continents. I also agree with the IDC quote that the two environments cannot “stray too far from the ER [entity-relationship] model”. This is because the data model defines the domain language, and the two environments must use a common language otherwise they cannot even communicate without an intermediary – that’s the antithesis of togetherness.
I had an interesting exposure to the upside of very loosely coupled development platforms last week. A major insurer uses this as a strategy for technology agility – the ability to morph not just business systems, but underlying technology as well. Their IT cost is a staggering low 1.7% of GWP – and they are large and growing. Their peers on more traditionally ‘integrated’ (assuming that this is a synonym for ‘coming together’) platforms are at multiples of this.
Regards
“You can see it here.” in my previous comment was a hyperlink that got lost. Sorry. Here is the link for those who are interested: http://sales.idiomsoftware.com/Idiom/idiomblog.nsf/dx/08052010063840p.m.MNO9T9.htm
Mark
Thanks for the comment. I believe that decisioning should be loosely coupled from data management (the CRUD applications discussed) and from process management/workflow. Decisions must be changeable independently of the contexts in which they are used. What I don’t think is helpful is the identification of lots of separate kinds of decision making technologies, all of which must be used independently. I think that rules and analytics, as well as optimization and yet-to-be-defined decision making technologies, need to be integrated enough that the right set can be used to solve each problem without having to install and configure a completely different product.
Like you I prefer decisions loosely coupled. If my post implied otherwise it should not have!
James, I tihnk I need to disagree with you about work on a decision management platform that unifies analytics and rules. Here at FICO we’re doing exactly that. We’ve gone beyond using PMML to export model information. Our analytic modeling tools now effectively share complete model information including characteristic calculations and data manipulations inside our Blaze Advisor rules repository. This goes well beyond what PMML is able to accomplish, through we still support PMML inport into Blaze Advisor.
Thinking about BPM and rules and analytics FICO has OEM relationships and partnerships that effectively create the kind of rules + BPM platforms that I think you are also describing. The future is now.
Mark
We are in violent agreement about the need to integrate the products (though I have yet to be briefed on what FICO is up to in this regard) and you are not alone – the Chordiant platform recently acquired by Pega, as well as newer products like Modellica (review coming) and Zementis ADAPA also allow rules and analytics to be effectively combined for decisioning. My problem is with the assertion that this is how these technologies will be sold in the future.
Frankly I just don’t see if for a while. Companies I speak to want to buy solutions that can only be built with a decisioning platform but they don’t yet, by and large, think of a decisioning platform as something they need. As I said:
It’s not the need for the platform or the value of a platform that is at issue here – I agree 100% – I just don’t see people buying it that way for a while.