Getting your AI partner right : 11 questions to ask before choosing your AI lending partner

Stay ahead of the curve with the world’s leading adaptive, regulatory compliant AI-powered Credit Decisioning Platform


AI in lending is still far from becoming mainstream. Credit scores invented many decades before modern computing, and AI's emergence often decides the fate of millions of members' dreams. Over 90% of all lending is still manual, less than 5% of loan decisions are instant, and even prime members pay too much for loans because the rates they pay effectively subsidize the losses from those who default.


AI is gaining ground with credit unions as a tool to enable more credit for members and avoid discrimination and bias. The use of AI in leveraging diverse data points to develop accurate predictive models that seek to identify creditworthy members is potentially a game-changer for the credit unions and can shore up approval rates and improve the member experience.


That said, multiple considerations must be kept in mind while evaluating the right AI partner for you. Using AI in credit decisioning can have numerous implications for you and your members. To understand which platform is right for you, credit unions must carefully assess their providers for fair lending compliance, expertise in risk management, and the accuracy of their models, among other factors.


With that in mind, we have curated a list of questions that every credit unions need to ask their AI partner to evaluate their competency. We have also tried to indicate how Scienaptic endeavors to tick the right boxes by staying true to the requirements of forward-thinking CLOs.


1

Is the platform designed to natively enhance the lending capabilities of your credit union?


Why does it matter? - Several AI vendors started as lenders. Their lending business failed hence they started selling the same technology which didn’t work for them or to get access to the deposits of credit unions and lend them on their behalf thereby depriving credit unions of natively enhancing their own capabilities.


Scienaptic - Scienaptic AI was founded with singular idea of enabling credit unions’ native capabilities so that they can help their members.

Scienaptic’s AI-powered credit decisioning platform is designed by seasoned credit risk & lending practitioners to address the needs of forward-thinking CLOs and VP Lending.



2

What is your vendor’s compliance strategy? What is their track record of adherence to compliance requirements?


Why does it matter? - Compliance is non-negotiable. The lending industry is bound by guidelines to ensure that there is no discrimination in lending practices.

Several AI vendors have been involved in rent–a–tribe, payday lending & other class action lawsuits. Google name of AI vendor you are considering + Lawsuit.


Scienaptic - Scienaptic’s AI models are natively designed to be explainable and ECOA & fair lending compliant. Compliance with all regulatory guidelines is part of our standard AI implementation program for every credit union. Our credit union clients have also gone through NCUA audit and found less disparate actions vs. human underwriting.



3

Are you getting just a custom score or are getting a full-service platform?


Why does it matter? - Many vendors provide custom scores only. CUs need to take that score and configure their strategies in the LOS using that custom score to the best of the abilities of that LOS and their own abilities to configure those strategies.

So, it is necessary to ask if the AI enablement in question is limited to just a scorecard? Or does it assist in the deployment, lending strategy, and rule execution as well?


Scienaptic - One unified flow for data connections, bestof-breed models, self-learning AI, rules, lending strategies, and ML Ops.

We set up rules and strategies (existing + new ones) and credit unions can edit/manage multiple strategies with ease.

The platform provides credit unions with the agility needed to dynamically react to market changes, rapidly test & deploy new credit strategies, update scorecards, and originate new loans with confidence.



4

Does your vendor’s platform make real-time decisions?


Why does it matter? - Most vendors have run batch processes with no real-time automation. Without a Business Rule Engine, there will be issues with legacy strategy/rule engines of LOS and the automation potential will be significantly limited, thereby making it impossible to make decisions in real-time, impacting member experience.

Ask your vendors if they have an inbuilt rule engine in place


Scienaptic - Auto-approve or auto-reject loan applications in milliseconds, creating a better member experience.

Our deep understanding of financial services and not just AI enables us to create scalable, seamless working of our AI scorecards with our strategy/rule engine



5

Would the vendor's platform support different credit products, including SMB lending?


Why does it matter? - Most AI platforms are limited to handling only a few credit products within consumer loans portfolio thereby requiring credit unions to find another AI solution for the remainder of the credit products in their portfolio.

It is necessary therefore to have a complete picture of the scope of implementation flexibility of the vendor technology across the spectrum of your offerings.


Scienaptic - Scienaptic platform can support varied consumer and SMB loan products. The platform is designed to handle complex workflows and data formats (e.g., cash flows, tax return filings, bureau features, liquidity ratios, macroeconomic parameters) that are inherent to any SMB lending process.



6

How integrable is the platform offering?

Will the integration cause disruption to existing processes?


Why does it matter? - Other AI vendors have one or two LOS integration. This limits their ability to integrate with your particular LOS.

Ask the vendor about successful LOS integrations with clients that are live on their platform. Try to get clients’ insights on how non-disruptive the process was.


Scienaptic - The platform deployment is designed to be LOS or Core agnostic. It works seamlessly with MeridianLink, Symitar, Fiserv, Temenos, Corelation, CU Answers, Allegro, Sync1, etc. For many of our clients, we have successfully integrated with their custom LOS as well.



7

Is the platform recession-proof?

How does the decision quality hold up in uncertain times?


Why does it matter? - It is imperative to take economic downturns into account and create balanced choices between growth and risk. Other vender models are not tested extensively for economic downturns, which can expose your portfolios to higher risk (e.g., a leading AI vendor is facing a class-action lawsuit for not adequately accounting for macroeconomic factors such as interest rates).


Scienaptic - Scienaptic's models are tested on past recessions and local calamity data. It helps you identify the most vulnerable members. It creates sharper differentiation between credit-worthy members facing hard times and high default risk applicants.



8

Will the platform help you get the most out of your existing rich relationship data?


Why does it matter? - Inquire about existing data partnerships that your vendor has. Most vendors lack a rich alt-data ecosystem that provides credit unions the flexibility to integrate a range of alt-data sources optimized for their portfolio requirements.


Scienaptic - We help credit unions get the most out of their existing member data, bureau data for which they are already paying and combine it with alt data relevant to their portfolio.

We have partnerships with all 3 bureaus, LexisNexis, 700Credit, MicroBilt, SentiLink, Plaid, CreditSnap, Urjanet, and Codat. Our platform helps drive financial inclusion by helping underserved, thin file or no file members to access to credit.



9

Do your vendor's model self-learn and adapt to your portfolio needs?


Why does it matter? - As one navigates in uncertain economic conditions, it is imperative that models adapt to your ever-changing requirements. You should be able to refresh your models as needed and always have the sharpest decisioning vectors at your disposal.

Unfortunately, most vendor models are static and do not auto-adapt to your everevolving landscape.


Scienaptic - Our adaptive AI factors in your portfolio needs, risk appetite, and hyperlocal context for delivering powerful signals.

After realizing initial gains, we keep the underwriting sharp by agile testing for predictiveness of new data, strategies, and scorecards.



10

Does the vendor platform deliver demonstrable results once live. Do they have published go-live press releases?


Why does it matter? - Impact delivered in a POC is vastly different from actual results in live production. There are many added complexities in a live production environment that make it hard to replicate POC success.

Most vendors make tall claims about business/member impact based on POC results but lack the tools needed to deploy it live with equal success


Scienaptic - All our reported business impact metrics are from actual live production clients / credit unions. On an average, most credit unions experience 15-40% more loans and 50-80% more automation in live production. Credit unions trying to contain credit losses see their bad rate drop by 8-15%.

These are not unverifiable claims. Many credit unions have gone on record through press releases to announce these gains.

You can read about these at: https://www.scienaptic.ai/news-events



11

Is the vendor platform suitable for credit unions of all sizes? Do they have small credit unions as their client.


Why does it matter? - Custom scorecards of most vendors work efficiently only when they are trained on large data sets of historical performance data that only a few large credit unions can provide.

This coupled with steep pricing that contains many hidden costs (e.g. periodic model refresh costs extra) excludes many small credit unions from the benefits of AI in lending.


Scienaptic - We are the only platform truly designed for diverse credit unions of all sizes. We have credit unions that range from $100MM in assets to $10BN in assets and everything in between.

In addition, we have auto-lending clients who are 1/10th the size of our smallest credit union client. This makes our platform well adapted and flexible for lending needs that diverse credit unions of different sizes may have.



Conclusion


The need for innovation in lending is inevitable, but so is the need for the right technology to fit your business and your vision. AI presents an unmissable opportunity to transform the dynamics of lending, but any credit union looking to take the leap of faith needs to do its homework. Only then can they see the real impact.

Scienaptic is your natural partner when it comes to AI. At Scienaptic, we have been at the forefront of implementing AI-powered lending for lenders, helping them reach more members, including underbanked and underserved individuals, and say “yes” more often without increasing risk. Our AI platform is used by credit unions of all sizes, integrating seamlessly with their existing systems.


182 views