DataCubes Blog

First Principles: The difference between “Prefill” and “Answers”

Posted by Phil Alampi on Aug 15, 2019 12:04:32 PM
Phil Alampi
Find me on:

We recently changed one of our product names from d3 Prefill™ to d3 Answers™. This was a very deliberate change based on the product’s core function to automate a critical point of decision-making within the commercial underwriting process. Both names leverage our concept of d3: Data. Discovery. Decision.®. So, what is the difference between “Prefill” and “Answers” then? To best explain, I need to go back in time to when I saw my first prefill demo:

The presenter logged into a typical insurance screen with about 50 data fields that need to be typed in – it was a normal experience for the time so far. Upon entering minimal identifying information on the risk, the screen came to life and nearly all the data points required were automatically populated! Wow!

It was a paradigm shift. It was 2007.

Since then, prefilling data points on a screen has worked so well and been done so often, it’s almost boring now. The message used to be “validation, not data entry,” which is all well and good – but no longer new and innovative.

Now we sit on the cusp of a new decade and we’ve found “prefill” does not fully capture what DataCubes does with this product. Sure we can prefill screens with any of our four billion data points on businesses, but there is much more to it than that.

 

prefillvsanswers

 

Why do screens need to be prefilled at all? Why do screens exist?

Getting to first principles, these types of screens only exist to capture information on the risk. Information only needs to be captured on the risk in order to facilitate underwriting and rating, so a quote can be provided. Therefore, if you can answer an underwriting question using AI, you don’t need to prefill a screen showing the data behind it. Instead, you can simply not ask that question, removing a good amount of the friction in getting a new business commercial P&C quote. The same logic applies for premium audits and renewals: we don’t need to ask questions we already have the answers to.

“But what about manual validation?” the pundits will say. Yes, that could be important in some situations. But do we need our users to do that in every case? If I have a permit showing a contractor worked above 30 feet or advertising from a restaurant showing they offer delivery, do we really need to cycle that back to the user in an already friction-laden underwriting process? It would be silly to do so.

 

datavsdecisions

 

d3 Answers is about getting to what matters about the risk and determining the best response to each underwriting question. Instead of a data dump, and thanks to AI trained on underwriting best practices, it cuts to the chase of what matters and allows human underwriters to focus on exceptions and agent relationships, where their talents are needed most. AI for decision automation - that is the real transformation happening right now and it will be a huge part of the next decade for insurance.