Berkeley Publisher – First Look

Berkeley Publisher is a product from Berkeley Bridge, a company based in the Netherlands. Founded by a group with experience in expert systems, especially in government systems, BerkeleyBridge has been focusing on “making knowledge profitable using intelligence software” since 2005, primarily in the legal domain. They are now focused completely on developing their platform as a horizontal capability aiming to integrate rule maintenance and execution components in existing or new architecture as part of a best-of-breed ecosystem.. Customers include universities and local government agencies as well as Allianz (insurance), essenscia (life sciences) and pwc (consulting).

Their target market includes many professional services companies, such as law firms and consulting firms, who see that quality, costs and billing rates are under pressure. Customers are increasingly unwilling to pay these organizations premium rates for very repeatable tasks (like NDAs or contracts in legal matters for instance). Similarly consumers are able to find more and more information for free but are overloaded with the amount of information – they cannot make decisions with it. In both cases one approach is for those with expertise to develop a system (a decision management system I would say) to handle the bulk of the cases intelligently while using experts to handle the more difficult cases.

While systems have been developed for this kind of thing before – the old expert systems platforms – they suffered from a number of challenges. Decision-making typically requires large numbers of ongoing changes and many more rules than was supported by these platforms(I call this the “rule management is more important than rule execution” problem). In addition communication between experts and the IT developers of these systems was a challenge (a platform for collaboration is key) and the resulting code was not easy to read of change (it lacked what I would call design transparency).

Berkeley Publisher is designed to address these challenges. The platform is designed to allow experts to share their knowledge in a decision-centric way that is scalable, uniform and transparent while giving them a way to focus their energy on the remaining, more complex cases.

The product’s core capability is to translate knowledge into decision trees and then manage and execute the rules represented by these trees. It has a graphical interface designed to support rapid prototyping so that users can begin quickly and evolve. The results can easily be published to a web service so that it can be shared over the internet. The end result is transparent, consistent, maintainable and expandable – all critical issues in Decision Management Systems. Example customers include permit issuing systems (which permits do I need), generating legal documents, determining compliance with HR regulations and assessing eligibility for benefits.

The basic management interface displays a graphical decision tree and a property editor for each node. The tree is actually a directed graph (to allow reuse) and can be managed as a set of graphs and sub-graphs to support broad re-use of core logic. Inputs – questions – can be defined for each node as well as document templates, calculations and other text elements. A set of actions is specified along with the conditions for those actions and conditions can also be specified when a sub graph is linked so that the graph is only called in certain conditions. Links to external sources (regulations, policies etc) is supported to help those maintaining the logic do so effectively.

In the interactive model, the tool will ask the questions, execute the subgraphs specified and then trigger the actions. This interface is generated directly from the specification using either a local engine (Berkeley Runner) or cloud publishing (Berkeley Webserver) and so can be used for rapid prototyping or interactive testing. The results can leverage the document templates so that the answers/decisions are plugged directly into a completed document.

The software can be integrated with other services using web services calls and the whole thing can be called programmatically with explicit inputs (from a database or form). This service version can run “headless” (returning a decision or a failure to decide) or it can generate an interactive interface when it needs to (to ask for missing data for instance).

You can get more information on Berkeley Publisher here and Berkeley will be one of the vendors in the next version of our Decision Management Systems Platform Technology Report.