I met General Intelletics at the recent Building Business Capability conference. The founder has been working in artificial intelligence and configurators for about 10 years. One of his clients wanted to put together an insurance program for a gold mine in Turkey – total premiums of about $4M/year. He found that there are no real tools to support collecting all the data involved and managing the program. These kind of controlled insurance programs have a lower limit today – anything less than about $40M are not profitable for specialist brokers to handle (too much senior time using traditional approaches) so the rest go to less skilled brokers instead. Plus there are folks like Oil companies who do many of these kinds of projects and want more automation.
PreSpec is the result – a product for managing the collection of this huge portfolio of information. It is focused on insurance today but clearly equally relevant to complex financial products, health care etc. The product is currently under development (I saw a prototype) and General Intelletics is look for a strategic partner.
PreSpec connects customers or clients with the insurance providers. There are 7 components to be built in several phases:
- Gather core facts (700-1000 facts perhaps) to get a basic quotation together
- Gather remainder of 3,500 facts to bind a quote
- Automatically create a standard client presentation from these facts for verification
- Expand fact gathering to third parties, experts etc to make information collection easier
- Decisioning engine to take these facts and pick the right coverages, explain why specific coverages don’t work etc.
- Links to a set of providers so they can automatically underwrite using their own systems when requested by PreSpec
- Allow end user customers to put the data in themselves for an end to end automated bound quotation.
The demonstration was of technology based on OpenRules.
Facts can be complete, recommended (by the software), possible issue, issue, available to answer or not required. Facts are grouped and the product is designed around different kinds of coverages though the underlying facts are shared. Different groups of facts can be selected and then completed. In the demo there are 14 facts and some 70 rules. Each fact has supporting documents (with a many to many cross reference), descriptions, help, audit trail with comments etc. As facts are completed they, and eventually their groups, turn green as complete. Typically 30% of facts are not required so managing this is important.
Facts are of different types and different widgets are provided to enter them. As facts are answered, up to two sub levels can be spawned. Facts also come and go in the interface based as answers to other questions make them relevant or not. Errors can be hard (a string when number expected) or soft (200 entered when a dollar value in millions is expected). Facts can be cross-checked against each other to flag inconsistencies. Problems ripple through the fact chains and groups, as do solutions.
In the end the value add comes in five ways:
- User interface for fact management
- Process for capturing the facts
- Explaining answers
- A more productive allocation of professional resources
- A simpler and easier client experience
I was intrigued by the product, which is addressing a clear need, and I look forward to seeing it develop.