AI for Sustainable Financial Inclusion

Lokyata helps clients create a

data strategy and execute it on
our AI-driven customer profile
analytics platform

Lending Decisions Predictive Analytics Customer Engagement

NOTICE: The contents of this document are CONFIDENTIAL and may not be shared with ANY third-party without the express, written consent of Lokyata Inc.
 About Lokyata

❖ AI powered API Platform

❖ Subscription-based Managed Service
❖ Easy integration using REST APIs with
✓ LMS, CBS, LOS & other systems
✓ 3rd party data sources

❖ Single Platform for

✓ AI-driven Lending Decisions
✓ Predictive Analytics
✓ Customer Engagement
 Lenders Mobile Operators eCommerce Platforms Wallet, Payments
and POS Providers
❖ Key technology for digital ❖ Increase checkouts and decrease
transformation. ❖ Monetize valuable telco
abandoned carts with prepaid ❖ Use payments data to
and digital wallet data provision credit to
❖ Increase efficiency, vouchers and instance credit.
providing a Mobile Score- customers.
accuracy, and volume with as-a-Service API.
AI-powered digital lending ❖ Access new customers who lack
❖ Monetize data through
decisions. credit cards or formal credit Wallet Score-as-a-Service
❖ Partner with banks to history.
❖ Implement new, fully provide customers and
❖ Increase engagement and
automated digital loan scoring for digital lending ❖ Analyze purchase data to generate reduce churn by offering
products. products, adding revenue product recommendations and loans on the platform.
❖ Integrate with existing and keeping people on the customer engagement messaging.
underwriting to increase platform. ❖ Use POS data to extend
volume and efficiency. credit to shop owners for
inventory or cashflow
❖ Lending, Savings, and smoothing.
❖ Generate messages from
AI engine to personalize Insurance increase
the user experience customer engagement ❖ Develop customer
and reduce churn behavior profiles to help
(GSMA 2018 Report). businesses engage their
customers and increase
payments transactions.
Success Story in Madagascar
Meet Nia Abara
❖ 30-year-old vegetable shop owner in Madagascar
❖ Needs funds to buy more inventory
❖ No banking or borrowing history

Nia gets her funds from a local bank!

❖ Nia sees a message on her phone for a new bank loan program and decides to try it!
❖ Lokyata gives loan decision to the bank. She is eligible for her first loan!
❖ In 2 minutes, Nia gets the funds she needs transferred to her mobile wallet.
❖ Today, Nia has applied for over 5 loans and is thinking of scaling her business.
Success Story in Madagascar
April 2018
Lokyata opens the door for underbanked
customers – with a Loan Decision,
Wallet Engagement and Analytics
Local Bank
Partner Partner

GOAL 24,000 <7% Portfolio Performance

Loans loss
5x
Superior to most popular
RESULTS 1,000,000 <3.5% program, M-Shwari
Loans loss

3.5 + 180 +
10 X Million Loans Scored 1.0 +
Million Loans Disbursed
Thousand Clients Impacted

Increase in clients within a year

Engagement Model
Data Assessment & Gap Analysis Data Strategy
Understand Client Business
Goals, Product & Target
Common Data Model
Segment ❖ Comprehensive set of customer and
business data points

❖ Standardizes APIs, Data, and Analytics

❖ Customers benefit from new features

and efficiencies developed for others

❖ Created from decades of expertise in

Lending and Inclusion space
 Traditional Scorecard Techniques Lokyata’s AI Driven Methodology
Credit Score Any and all Automated AI
available data Comprehensive Development Pipeline Automated, Objective &
data framework Transparent Decisions

S tat i st i ca l Re g re s s i o n
Variables 1
Variables 1…1…
Variables nn
Variables 1…

Common Data Model

Financial

Decision Codes
TransparentAI
Variables 2 Credit Score
Variables
Variables1…
1…nn
Bureau Variables 1… n Sub Scores
Grantable Amount
Factor Multiplier Risk-based Loan Features
699.45
Variables 3
Variables 1… n
Variables
Variables1…
1…nn
Industry/sector

Variables 4
Variables
Variables1… n n
Variables1…
1… n
other

1000s of features created from existing raw variables |Automated feature creation| Train multiple models and evaluate
each one| Select the best performing model given the available data and business goals| Monitor & finetune models
Executed on Lokyata’s API-First Platform

Customer Engagement & Customer Profile & Lender Portfolio

Messaging Score Analytics

Customer Lender
What’s Unique?
 Creates Multidimensional Customer Profiles

Demographic Economic
• Identifiers (IDs, DOB, etc.) • Macro
• Personal (Gender, Marital • Micro
status, etc.) • Geopolitical
• Education & literacy • Industry/Sector
• Location • Commodities
• Forex/Currency

Non-Credit Transaction Environmental

• Wallet Transactions (Cash in/out, • Weather
remittances, p2p, etc.) • GIS
• Purchases • Agriculture
• Utility payments • Satellite
• Card Usage/transactions
• UPI Transactions
Financial Device
• Personal • Telco
• Household • Phone Call / SMS History
• Business • Location Services
• Social Media Data
Credit / Saving History • Internet Usage
• Requested loan facility (behavior) • Installed Apps
• Internal to Lender • NLP and Images
• Third party
• Credit Bureau
• Guarantees / contingent obligations

Comprehensive & Unique Profile with any all available data

Comprehensive Customer Profile

Borrower Score
746.4/1,000
B9 Credit Behaviour
Outcome (Score)
430/1,000

B9 Credit Behaviour
Outcome

B9.1 Payment History

B9.2 Repayment Behaviour

❖ Multi-dimensional Customer Profile created using any and all data available on a customer
❖ AI-driven methodology enables identifying all possible relationships between variables to create a comprehensive data profile
Decision Codes

✓ Transparent decision
codes at a customer
loan level
✓ Allows comparison of
customer profiles to
similar profiles to
ensure that decisions
are more accurate and
reliable
✓ Visibility on factors that
impact a customer
profile negatively or
positively to enable
customized customer
engagement
AI Scoring is only one element of the lending decision

Data Scoring Model Business Rules Policies

• Generate Multidimensional • Grantable Amount • Accomplish any desired

Customer Profile and outcomes explicitly
• Fraud Prevention
overall risk score (e.g. No GA reduction
• Min/Max Loan Amount for good borrowers)
• Machine Learning Driven
• Interest Rate • Business Goals Driven
• Fully Automated
Development and Training • Business Goals driven
Explainable AI (XAI) – Not a Blackbox

Enables Responsible and Fair Lending

Demonstrates how the model makes

decisions

Validate through Integrated Reporting &

Analytics

Model Level & Loan Level Transparency

 Single Platform for all Loan Products
Agri MSME

Digital Micro
Predictive Analytics
Powered by AI
Track historical performance and use Lokyata’s financial modeling to forecast Track custom financial inclusion metrics which are most important to you directly
future performance under multiple scenarios through customer data (i.e. digital payments)
Track fair lending outcomes. Shown here is an
application distribution and approval rate
distribution for various demographic categories

Results which may raise concern are

automatically flagged & can raise real-time alerts
AI for better decisions!

01 02 03
AI algorithms perform Predictive AI based models have already been
AI approaches are dynamic and analysis. Models are constantly perfected and are the way forward.
objective. fine-tuned to accurately predict Early adoption is a strategic
Identifies new insights from outcomes on newer market decision that will provide
existing data. segments or geographies. competitive edge !

04 05 06
The benefits of an AI driven model Lokyata designs Custom AI/ML models look at thousands of
far outweigh the investment. Models and Data Strategy with inter relations and provide
Lokyata provides its AI platform at a as little data as available objective decisions. Start with any
low upfront cost and a pay as you utilizing our Expert → Hybrid data and build and improve from
grow subscription model. real-time data and outcomes.
→ ML model design approach.
Simplifying the Journey to Launch!
1

3 5

2 Raw Data
8 Models
Schemas
2 APIs
2 Launch
Weeks Weeks Weeks Weeks

4
2
Thank You!