Understanding your legacy risks
You are the chief underwriting officer of a multinational reinsurance firm. You have just been asked by your business leadership to quantify your organization's global property and excess liabilities as part of an across-the-board evaluation of broking standards, exposures and underwriting guidelines. You send your staff off to crunch the numbers from contract data in your data mart, yet you know that any analysis you do is skewed to the point of being useless in its predictive capabilities. This is because you most likely do not have access to much of the data within these contracts. Answers to questions like "how much liability it tied up in aggregate vs. per event, across geography and cedants going back 10 years?” This data is locked up in paper. In layers of contracts across multiple years from multiple jurisdictions. All of which requires a level of painstaking review to extract the data within them into a usable format.
Until today, this activity was expensive and error prone, requiring expert human review and oversight of countless documents. This is typically not a good use of high value underwriter resources, whose time is valuable and better used elsewhere. Clearly a better way has to be found.
Cognitive computing tools such as Exari Vision are now available to help; with tools that would first provide a semantic classification and clustering of relevant clauses and terms within these legacy contracts. In addition to these traditional language processing algorithms, advances in artificial intelligence that uses neural network technology now has the ability to detect anomalies in patterns of text that are very useful in highlighting potential problems in contracts; a task that is very difficult for humans to conduct, especially at scale. As this new technology becomes available to assist in business decision making, your organization will have the ability to focus resources on fixing issues with legacy contracts. The next stage of analysis where cognitive technologies can then help is in applying predictive models to the data that has now been extracted from the documents. You can begin to answer questions around cedant and broker behavior, predict trends within geographic areas and other business scenarios. You can start to see how this technology can be used at various stages of your journey towards unlocking information from paper contracts.
You open up a number of new business opportunities, and there are now a variety of cognitive tools available to help your organization at multiple steps of the way. Each of these tools are designed for specific problems, and we at Exari recommend using the right set of cognitive tools in an evolutionary process from initial description to predictive modeling as a best practice. We start with statistical and genetic algorithms that describe the data in your contracts. The next step is diagnosis, with machine-extracted classification technology to help segregate good terms from bad ones, and Neural nets trained to spot anomalies across individual contracts. Finally, we get to the holy grail of predictive modeling once the data, terms and clauses from your contracts are completely digitized and available to your business and data sciences teams.
With Exari, you finally have the necessary tools to go back and quantify your organization's global property and excess liabilities with predictive analytics.