Helping lenders make better decisions through artificial intelligence and novel use of data.


Business financing

Payment processors, mobile carriers and even e-commerce providers are jumping into customer financing. While this opens up growth and revenue opportunities, it can also bring massive credit related losses.

Our A.I. doesn't stop at underwriting. It crafts custom credit offers and estimates necessary reserves.

Small Business Loans

Institutions with less than $10 billion in assets provide nearly 60% of small-business loans.

Community banks and credit unions have a unique opportunity to be part of the second fintech wave by applying our A.I. technology to internal and third party data sources.


Non-conforming loans comprise 36% of the mortgage industry. Banks that offer jumbo, Alt-A, A-minus, or subprime loans leverage our machine learning models to improve default, delinquency and pre-payment predictions.

With our unique approach, banks don't have to choose between predictive power and model transparency. 

Changing consumer behavior is being joined by new technologies to drive a new era of innovation in financial services.
— Goldman Sachs Global Investment Research

Understanding the Borrower

Our models include a wide range of summary characteristics about the borrower like credit score, loan-to-value ratio, etc. However we also use detailed income and expense data including credit card transactions, bill payment history, accounts payables and receivables, outstanding loans and investments.


Deep Learning - Property Images

Property images contain a wealth of additional information helpful to valuation models including building material quality, amount of sunlight, floor plan layouts, etc.

We use the latest computer vision deep learning techniques to incorporate property image data into underwriting models.

Upgraded home with hardwood floors, upstairs & downstairs decks for entertainment & views of the valley, hills and reservoir. Sitting on .70 acres, close to downtown and theater square. Great commute location!

Deep Learning - UNSTRUCTURED text

A significant portion of the data in the underwriting process can be unstructured text. Traditionally attribute extraction is done against such text to structure it for analysis.

By using a deep learning approach, we can directly feed unstructured text as model inputs. This ensures that more of the signal in the text is preserved and eliminates the need for rules-based attribute extraction.