Your browser is not supported. Please update it.

8 June 2018

Black Knight buys into AI for mortgage approvals with Heavy Water

Since the credit crunch a decade ago, banks have faced increasing workloads approving mortgages because they have to employ tougher checks against mis-lending and toxic debt. At last, AI techniques are helping reduce the cost and time of mortgage approvals while inevitably raising the question of whether jobs in the financial sector are in jeopardy as a result. For this reason, Black Knight, a firm based in Florida, analyzing mortgage applications for large banks, was keen to emphasis that its move into AI will displace its staff into more creative roles rather than replace them.

Like many other established firms across a range of sectors, Black Knight has fast-tracked its recruitment of AI through acquisition, buying a two-year-old start up called Heavy Water whose AI product Aiva (Artificial Intelligence Virtual Assistant) is still under development. Even so, the choice was quite easy for Black Knight because Aiva was created specifically to accelerate loan approvals with mortgage lending the first target. Black Knight said it will now direct the remaining development of Aiva towards deployment as a product it can offer directly to its clients, while enhancing its overall offering which currently generates sales of $1 billion a year.

Black Knight was attracted in particular by the fact that Heavy Water’s founder Soofi Safavi came from the mortgage industry and was inspired by frustration with expensive IT systems developed for his employers. He served at JPMorgan Chase’s mortgage division in Edison, New Jersey before joining Radian, the Philadelphia mortgage insurer, as CTO until September 2015 when he finally left the industry to set up Heavy Water. He cited lack of innovation despite the incentive of spiraling costs in the post credit crunch era, which of course created business for consultancies such as Black Knight.

Safavi estimated the cost of starting a new mortgage at around $2,500 in 2007 just before the credit crunch but that has soared to over $8500 today, which has driven smaller lenders to consolidate so that they have the scale to absorb the costs. IT so far has merely contained and managed the problem rather than seriously reduced the cost and complexity, according to Safavi, which is why he saw an opening for a start-up. Black Knight was not even prepared to wait for Aiva to be proven which indicated it saw an urgent need to streamline its own processes in helping its client lenders approve mortgages.

Mortgage lending is still a huge business in the US with total mortgage debt running at just over $14.5 trillion – so it’s hardly surprisingly there are plenty of other AI start-ups in that field. There are also larger scale collaborations, with some even getting into the loan market directly. One of these is ZestFinance, which specialized in online lending to people on the edge of credit worthiness previously unable to secure mortgages, founded in 2015 by ex-Google CIO Douglas Merrill. This launched a loan service called Basix for this purpose, highlighting how machine-learning can generate new business for lenders as well as cut costs.

Black Knight is also interested in taking Heavy Water’s Aiva in that direction once it has demonstrated its capability for absorbing complexity. It is the ultimate big data problem, relating an individual’s creditworthiness to data from many sources, including the obvious ones like value of collateral such as cars, homes, businesses and works of art, taking in factors such as predicted levels of inflation and growth in value of the asset concerned, a house.

But as Black Knight stressed, machine learning can fine tune this process and also make much more accurate assessments of risk by including additional sources and adapting weights on the basis of past outcomes. It may find correlations between post codes and nature of employment that help calibrate the risk and this opens up the potential for lending to people previously denied loans, while also reducing levels of defaulting by not lending to those at higher risk.

The same software can also be readily shoehorned to other types of loan involving enterprises, which Black Knight is hoping to exploit. Heavy Water is based at the University City Science Center (UCSC) in Philadelphia, the first and largest urban research park in the US.