In 2014, John R. Coyne developed a Semantic Armature for regulatory compliance filings that was introduced to the market through the highly regulated insurance industry.
The centerpiece of his Armature product rollout coincided with the advent of post-financial crisis mandated enterprise risk reporting as prescribed by the National Association of Insurance Commissioners (NAIC), which typically promulgates model laws and regulations on a wide range of insurance and related issues, including the extensive process of insurance holding company reporting.
Mr. Coyne’s vision for his Armature anticipated that the adoption of these rules would come on a state-by-state level, meaning nuance was bound to prevail. And so it did. With each state making respective adjustments to its adoption of the NAIC’s models, these resulting nuances now lead to differences of a substantive nature in filing requirements and added complexities of multiple state filings that compliance officers had not seen coming.
Mr. Coyne’s answer was to develop a regulatory alignment among all the states, resulting in a base knowledge model that obtains all their respective necessary processes. The next model to interact with it that in the Armature is the actual rules of filing, followed by a knowledge model of what regulators actually mean by their questions. Mr. Coyne’s Armature also includes a model of best practices to advise users on the best outcome for a particular filing.
By basing his Semantic Armature system on this regulatory alignment, its inference engines (based on artificial intelligence or expert systems techniques) can identify and manage the complexity of multi-state filings. Why is that important?
Insofar as the Armature’s use in the insurance industry, for example, for a small domestic insurer filing in a single state, the complexity is low. But for an insurer with multiple operating and risk bearing companies in its network, the complexity can be so onerous, that an acceptable filing using manual techniques would be otherwise almost impossible without an army of legal, regulatory and compliance professionals.
In the insurance industry application, Mr. Coyne’s Semantic Armature system automatically determines the business rules from analysis of a network of companies in an insurance holding company system:
- WHO (investors and/or assignees)
- WHAT (informational)
- WHERE (per state with the nuance)
- HOW (methods and best practices) and
- WHEN to file.
The “magic” in Mr. Coyne’s armature creation is the separation of concerns that semantic technologies allows. It also incorporates the best features of network and hierarchical databases with current triple models like RDF. By defining the models in a declarative way, WHAT is required can actually be separated in computational terms from HOW to find the requirement.
Separation of What from How
In the simplest of terms, one of Mr. Coyne’s legacy transformation collaborators, Don Estes, estimates that, in normal code development, roughly 70% of the code is telling the application how to perform its function, and what resources it needs (networks, data, other applications), so only 30% of the process is dealing with an actual business rule. Mr. Coyne’s Armature separates that and allows a specialized inference engine to figure out the HOW.
Nevertheless, all that concerns a user are the outcomes. Mr. Coyne’s Armature’s logic models are therefore free from infrastructure and highly transportable, thereby reducing the potential to become legacy while improving agility. His system is designed to change business rules, statutes, regulations and advisory knowledge with zero negative impact in less than a day.
In addition, the context is always maintained so that the separation of concerns also naturally flows to the users of the system. Only those with specific need know and file are permitted to view or use certain aspects of the system. This allows multiple collaborators to use the system in a discrete way.
Users as part of the system
Another major difference in Mr. Coyne’s Armature’s design is that users are a part of its system and play a functional role in processing. Subject matter experts interact with the Armature’s environment and become the biological code within its total system. What this means really is that the human factors allow the system to learn, adapt and change rapidly due to the design of incorporation of humans as part of the system. A value added is that, as the system learns and adapts, straight-through processing is increased over time.
Impact Assessment planning and avoidance
Another feature that falls out of the design is what Mr. Coyne terms “Future Casting ©” Because the inference mechanisms are so robust, users are able to test the impact of potential changes in statutes, regulations and business practices well in advance of their actual impact, so that their businesses can prepare for how they are going to respond. Armature users can now develop a plan of action, rather than simply react—which is largely the current practice.
Anecdotally, as Mr. Coyne’s team was testing the knowledge models under his direction, his Armature system autonomously discovered that one of the states in his test had a regulation for which it had no statute. Mr. Coyne directed his team to contact the corresponding regulator, who agreed that the regulation would indeed have to be removed from use and new legislation proposed to correct the anomaly That sort of logic error would not have been discovered conventionally unless a subject matter expert were looking for it.
Economics of development
From a macro business perspective, this technological approach is less expensive to develop, can be non-invasively and non-destructively integrated into current systems and provide Real Time Regulatory Oversight rather than remediation.
From a cost point of view, one of Mr. Coyne’s collaborators at Red Pill Systems, (www.redpillsystems.com) produced an analysis that determined that 100 simple declarative statements replaced nearly 150,000 lines of code.
Mr. Coyne estimates that his Semantic Armature development comprises at least 500 statements of these kinds (concept level abstractions) and therefore equates to 750,000 lines of code. At a normative cost for a full test implementation, a line of code costs around $30; equating to roughly $20MM of development. Based on normal development cycles, this would otherwise take a team of roughly 40 IT staff members about a year to develop and implement.
After one month of solid design effort (building the logical armature) Mr. Coyne began development of this system with one lead architect (semantic modeler) driving process that fell out of the logical structure, one support person developing supporting knowledge models and two lawyers trained to interpret statutes and regulations into knowledge models. The result was a completed system in four months—the equivalent of $20MM worth of effort for under $1MM.
Ready for any regulation
Of equal importance is that Mr. Coyne’s Semantic Armature system is now ready for the adoption of any regulatory change.
In the insurance industry, for example, a compliance requirement known as the Own Risk Self Assessment (ORSA) will be adopted by 2016 and, although replete with quantitative analysis, the qualitative application of ORSA’s rule sets will be easily adopted into Mr. Coyne’s system, be able to surface conflicts and anomalies and be ready to be fully aligned with all statutory compliance requirements.
Because of the robust nature of Mr. Coyne’s original design and architecture, his Semantic Armature can be adopted to any regulatory space, including banking, healthcare, environmental or any of the myriad new regulatory authorities covering everything from the cradle to the grave.
Mr. Coyne’s system is currently known as ORCA for Open Regulatory Compliance Armature and is due for release in 2015.