HALO Empowering Lenders with Artificial IntelligenceVasudevanSwaminath - PDF document

HALO Empowering Lenders with Artificial
HALO Empowering Lenders with Artificial IntelligenceVasudevanSwaminathanPresident and Principal ConsultantWhy HALO?OUR SOLUTIONHALOA machine learning based scoring model that helps in eliminating bad leads and approve good leads based on past dataHelps lenders fund the right merchants and less of the wrong ones Consumes “Big Data” that includes credit reports, bank statements and other dataBuilds a scorecard that self trains and improves based on a number o

f factors��2&#x/BBo;&#xx [
f factors��2&#x/BBo;&#xx [1;.63;X 7;&#x.331;
12;.16;” 2;.35;' ];&#x/Sub;&#xtype;&#x /Fo;&#xoter;&#x /Ty;&#xpe /;&#xPagi;&#xnati;&#xon 0;&#x/BBo;&#xx [1;.63;X 7;&#x.331;
12;.16;” 2;.35;' ];&#x/Sub;&#xtype;&#x /Fo;&#xoter;&#x /Ty;&#xpe /;&#xPagi;&#xnati;&#xon 0;www.zucisystems.comLENDER CHALLENGESUnderwritingUnlike large financial institutions, a major challenge for lendersMost lenders say overall los

ses within acceptable range but wants to
ses within acceptable range but wants to price the risk at the top of the funnel to grow the market.As a result they:Buy leads and make loans for applications they shouldn’tPass on leads for applicants they shouldHALO What it offersLead Rejection AccuracyLead Selection AccuracyIncrease in more creditworthy borrowersReduction in losses and default ratesHALOEURISTICALLYPROGRAMMEDGORITHMICUTPUTA 360* SOLUTION FOR YOUR UNDERWRITING NEEDSwww.zucisystems.comSam

ple DataAll this contribute in determini
ple DataAll this contribute in determining the actual payment performance of the merchant to whom the loan was issuedAttributes received from the alternative credit score (Inputs from other systemsApplication SourceNumber of times same “lead” encountered Inputs selfprovided by the applicant (Comparison with the actual data lenders get on credit reports & bank statements etc.)Demographics (Geography, Industry, Age, etc.)Attributes received from fraud scoreC

ash flow and transactions from applicant
ash flow and transactions from applicants bank statements (Decision Logic and/or other APIs)Choice the applicant makes as to amount, term and paymentType of bank account (Business or Personal)Type of business entity (LLC, Corp, Sole Prop etc.)Issue Time of Business Tax IDSCORECARDSLead to Prospect( Algorithm 1)Prospect to Customer ( Algorithm 2)GoodPoorPoorSA Strong AcceptAcceptReviewSR Strong RejectGAN (Generative Adversarial Network) : Outcomes Expectati

onAlgorithm 1 (Lead to Prospect) Outco
onAlgorithm 1 (Lead to Prospect) OutcomesApprovedPendingCollectionDeclinedWithdrawnZuciAlgorithm Result 1000 CustomersOutcomeAcceptReviewRejectHALO HOW IT WORKSPOWER YOUR LEAD AND UNDERWRITING WORKFLOWHeuristically Programmed Algorithmic Output (HALO) helps financial institutions in improving their lead identification and underwriting efficiency process.Improvement in Underwriting Efficiency��40%&#x/MCI; 24;&#x 000;&#x/MCI; 24;&#x 000;ww

w.zucisystems.comLearning Based (NonRule
w.zucisystems.comLearning Based (NonRuleDecision is not rule based, thus making system adapt to changing situations over a periodThird Party Integrations:Can establish connections to any thirdparty data sourcesContinuous Learning Models: Models are generated with change in customer and decision data in an unsupervised fashionInstant Processing: Decisions are generated realtime or batch processes, with comprehensive audit trailsAPI Based Integration: Can easily connect

to underwriting workflows or LMS / LOS
to underwriting workflows or LMS / LOS Systems via API endpointsHALO HOW IT WORKSHALO Other Features Provides Handyman features for all Datacentric problems, be it configuration, integration or management consoleConfigurationAccess To Multiple DataTime SaverOptimizes ROIManages DataData Friendly EnvironmentManagement ConsoleOne Stop Key SolutionNative Language SupportHALO Implementation Timelines and OutcomesData AnalysisData Sourcing Data Cleansing Data AnalysisF

eature ExtractionParameters ReviewParame
eature ExtractionParameters ReviewParameters ConsiderationParameters SelectionScreening AutomationApplication Screening RulesRules EngineAutomate Application ScreeningTrainingCreate ModelSet Score ThresholdGenerate predictionsConsumptionConsumption using API/UIAnalytics on overall efficiency0 -0304 -0708 -1112 -15Data AnalysisWeeksScreening AutomationTrainingConsumptionFeature ExtractionHALO Milestone based Pricing ModelTotal Cost $ 36,000 Hosting Charges, Scope

Changes, Other Charges 20% of Total Co
Changes, Other Charges 20% of Total CostTimeframeEnd of 3 weeks$ 7,20040% of Total CostFeature Engineering ShowcaseTimeframeEnd of 7 weeks$ 7,20080% of Total CostTimeframeEnd of 15 weeks$ 14,400100% of Total CostTimeframeEnd of 19 weeks$ 7,200Competitor Analysis 40% Better Cost l Improved EfficiencyCustomerTestimonialAt its core, Zuci'sunderwriting solution HALO (Heuristically programmed Algorithmic Output) helps us in two areas eliminating bad loan applica

nts, and identifying good loan candidate
nts, and identifying good loan candidates.The HALO solution is set up to learn on its own, without the need for manual adjustment to the rules. Zuci'steam built this model based on lead, applicant and consumer historical data with the ability to selftrain and retrain itself based on any updated data received by the system.As Zuciexplained, this system uses a Generative Adversarial Network, a class of machine learning systems that has helped significantly improve lead

rejection accuracy and lead selection ac
rejection accuracy and lead selection accuracy within 6 months of implementation. We are confident that HALO will continue to provide us with significant improvements over time.James C. JacobsonPresident at First Financial Service Center“11Thank You6912 Main St Suite 227 Downer’s Grove Chicago, IllinoisUS: +1 (312) 774India: +91 (44) 49525020sales@zucisystems.comThe picture can't be displayed.The picture can't be displayed.The picture can't be displaye