Regional bank leverages artificial intelligence to improve prediction rate
Success story snapshot
99.36%
Jewel Loan
Random forest classifier, gradient boost regression
93.00%
OD Enhancement
Random forest classifier, gradient boost regression
96.33%
Churn Prediction
Multi outcome random forest classifier
97.00%
Loan default prediction
Reasoning cluster model
98.00%
Product recommendation
Multi outcome random forest classifier
97.00%
Loan default prediction
Gradient boost classifier/K-means clustering
Our management board had their reservations before the engagement. But we were pleasantly surprised seeing the results from the initial phase. Intics.AI team hand held our internal team members to define scope of work, create AI strategy, train internal team members & communicate success metrics to everyone.
– Chief Digital Officer
High level solution architecture
| MIS reports capital generation drill down – asset quality improvement – lifetime value enhancements – compliance status – churn rate – use case specific reports | ||
| Storyboard sandbox |  | Approved use cases |
| Intelligent caching | ||
| Prediction engine middleware components default risk detection – churn prediction – lifetime value – upsell – cross sell – retention – text extraction – tabulation – attribute key value – information extraction – document and data classification – image detection – asset classification – charts – graphs – visual color codes – hyper-personalization templates – report templates – communication templates | ||
| Demographic data | Transactional data | Other data |
At Intics.AI, our team cares deeply about the environment and society at large. We measure carbon footprint at every stage of data processing and optimise emissions through intelligent caching. We strive hard to help institutions achieve their carbon credit goals.
Anilkumar Sanareddy
Intics.AI
1
Storyboard Sandbox
Senior business users can create use cases,
simulate data lakes and backtest strategies
2
Approved Use Cases
Run time business users access and connect to
relevant prediction engine components
3
Integration
Connectors and adaptors to internal & external
data sources and front end applications
4
Intelligent caching
Access to relevant data cohorts based
on approved use cases
About the customer: Regional bank based in South India. Their key revenue segments include interest & discount on advances & bills, income from investment, interest on balances with RBI and other Inter-Bank funds interest.
● Intense competition from fintech upstarts, lending institutions & legacy banks
● Conservative culture with reservations about digital technology adoption
● Lack of comprehension about unit economics beyond senior management
● Stringent IT policies which complicated maintenance processes
Solution: Intics.AI’s deep learning solution comprehended the seemingly chaotic and complex behaviour of the bank’s consumer base spanning multiple generations based on their distinct needs and diverse wants.
Highlights: Our deep learning solution created a neural network training model based on 150+ consumer demographic data, transactional information and behavioural aspects.
This holistic approach helped the bank:
- Generate risk score to validate credit worthiness and create early warning systems of defaulters
- Provide intuitive and timely nudges to bank executives which help increase upsell and cross sell offerings
- Create hyper-personalised messages to revitalise passive consumers
- Identify gold mines of unstructured data and digitise them. Eg: Extraction of text data from scanned copies of AADHAR card
- Identify gold mines of unstructured data and digitise them. Eg: Extraction of text data from scanned copies of AADHAR card
- PAN card, driving licence, business contracts, collateral hypothecation agreements
2150 S Central Expressway, Ste 259, McKinney, TX 75070
