There are four demos on this web site. They have been selected and developed to show some of the variations possible with policy components, their efficiency with computer resource, and increasing complexity. This page provides a brief description of each of the demos and links to each one.
This is the simplest of the demos logically and functionally. It takes an insurance policy number and attempts to assess what the payee code for that policy might be. There are no straightforward answers and no slam dunk successes. This system fits into the real life work flow of an insurance company which processes millions of transactions a month on their own and on their clients' behalf. The clients are mortgage companies; the payee codes relate to the office from or to which the transactions pertain.
A description of this system has also been published by The Modeling Agency Newsletter for July 2006: http://www.the-modeling-agency.com/news/2006q3.html
This demo automates an utterly unremarkable mortgage banker's price sheet. There are several variables involving: credit history, loan characteristics, property characteristics. This sort of price sheet has long been the subject of "automation" with rule-based tools and environments. The intent here is to show not merely that this sort of automation is possible and more efficient with a policy component, but also that the price sheet has some inflexibility in its pricing which can be improved upon with a policy component, thereby improving the bank's competitive position.
This is purely hypothetical and simplified from real life insurance, but contains most of the components of driver and automobile attributes. This demo is the first in the list to illustrate full expert system functionality, although in a simpler form than one might ordinarily expect. Not only is the functionality there, but so is what may to experienced aficionados appear to be rather phenomenal efficiency.
This is the clincher. This demo parses a text-based credit report and scores it with multi-dimensional attributes. Subsidiary scores deal with specialized factors such as reason codes, mortgage history, etc. This demo is nearly ready for prime time to evaluate loan applications, to give heads up to servicing as necessary, to score entire portfolios, or even to build loan pools for Capital Markets.