Data Studio AI - driven Data - as - a - Service

Challenges in Enterprise Data Management

DaaS simplifies a multi - step, complex journey Typical Enterprise data flow Complex. Unreliable. Time - consuming. Onboard Monitor Process Enrich Analyze Execute • Most enterprises agree that more than 40% of their time has been spent on integrating data across systems • With anywhere between 20 to 200 systems in a typical enterprise, 60% of executives feel they cannot trust the data and insights they have • Globally, 81% of fintechs say data issues are their biggest challenge. ✓ comes pre - built with access to external data sources, AI dashboards and models serving use - cases ✓ simplifies data management, quality and accelerates implementation of AI - led use - cases/apps ✓ reduces time to insight; guides enterprise buyers, using AI DaaS Data Platform Seamless. Intuitive. Intelligent.

Relevance of AI driven DaaS – Distributed Information • With anywhere between 20 to a 200 systems in a typical enterprise, 60% of executives feel they cannot trust the data and insights they have • Most enterprises agree that more than 40% of their time has been spent on integrating data across systems • For the ones that do start - the average time to get data - driven insights is around 3 - 5 years involving hundreds of people and dozens of departments. 1

Relevance of AI driven DaaS – Visibility to Right Dataset • Data is not accessible to all stakeholders. • 71% of banks underperform at collecting and using customer data. Only 8% of banks can apply predictive insights to make decisions and run campaigns. Even digitally savvy fintech companies struggle in this space. • Globally, 81% of fintechs say data issues are their biggest challenge . 2

Relevance of AI driven DaaS – Quality of Data • Available data is of poor quality. • Inaccurate data costs enterprises 15% to 25% of revenue. “Dirty data” costs the global banking industry over 400 Bn USD annually 3

Relevance of AI driven DaaS – Siloed Data • Data available in Silos • Business units create their own siloed data and process it independently. This process could prove to be inefficient when we deal with heavy data. Ultimately, making it difficult to present the same holistically. 4

Relevance of AI driven DaaS – Integrated Data (External & Internal) • External data is not used • Organizations that use external data saw 37% more revenue per employee • 92% of analytics professionals say their firms need to increase their use of external data sources • Only 28% of banks can integrate structured customer data to use in AI - led initiatives. There is a skewed view of the customer’s lifestyle and financial preferences 5

What sets Crayon’s Data Studio (DaaS) apart

AI driven DaaS - Pillars Seamless onboarding • Automated mapping technology • Ready - to - go connectors and data loaders • Intuitive data onboarding experience Enrichment and discoverability • GPT - powered enrichment • Natural language querying for dashboards, models, insights Automated data quality monitoring • Pre - built business rules • Easy addition of new business rules • Reporting and analytics on data quality Pre - trained AI models & dashboards • Pre - built models for every business domain • Ready - to - go dashboards to provide critical insights • Rich metric and feature stores for additional modeling and signal detection

Relevance of AI driven DaaS – Success Metrics Insights into performance on target and what should be actioned to achieve the same Increase in Customer Lifetime Value, Customer Acquisition/ Experience Product performance The most effective DaaS capabilities take a multi - pronged approach

Data Lakehouse

Key Component of DaaS - Lakehouse Teams Siloed layers driven by use cases Use Cases Data Views Data Lake Raw tables Source Data table 1. Business 1a 1b 1c Core Banking Card Management 2. Marketing 2a 2b 2c Loan Booking Branch Banking 1. Risk & audit 3a 3b 3c Digital Bank Mobile Banking Move From Siloed layers driven by use cases To a unified ML/Analytics data layer • Provides best of both worlds! Lower cost of storage, accommodating diverse data formats while holding to the principles of being Atomic, Consistent & Durable . Lakehouse – Unified analytics data layer with Standardization, Governance

AI Driven DaaS - Data Lakehouse - Design • Gain valuable insights from the data, drive innovation, and stay competitive in today's business landscape • Includes components such as object stores, data layers, processing layers, semantic layers, communication layers, client layers • Seamless data modeling and processing • Reduced Data movement and redundancy • Real Time updates • Schema Evolution – Schema support with mechanisms for data governance • Store structured, semi - structured, and unstructured data Simplicity Flexibility Volume Volume Volume BI Reports Data Warehouse ETL Structured Data Data Warehouses BI Reports Data Science Machine Learning Data Lake Data Warehouses ETL Data Lake Structured, Semi - structured & Unstructured Data BI Reports Data Science Machine Learning Data Lakehouse Metadata and Governance Layer Data Lake Structured, Semi - structured & Unstructured Data Time Breadth Breadth Time

Canonical Modeling • Building a canonical and robust data model is crucial for organizing and representing data effectively. Here are some key steps and considerations to help you in this process: Understand the Business Requirements Conduct Data Analysis Define Entities and Relationships Normalize Data Define Attributes and Data Types Handle Data Hierarchies Document the Data Model Iterate and Validate Consider Industry Standards Maintain and Evolve the Data Model 1 2 3 4 5 6 7 8 9 10 By following these steps and considering the unique needs of your domain, you can build a canonical and robust data model tha t e ffectively represents and organizes your data for analysis, integration, and application development purposes.

Data Studio simplifies the data journey

Simplifying complexity in enterprise data flow in one seamless flow Onboard Monitor Process Enrich Analyze Execute • Capture Source Information from the UI • Map to data Dictionary • Store Source details • Store Audit details • Run data ingestion pipelines for RDBMS, flat files, JSON, and XML • Identify the mapping (Source to Target) with name matching, metadata matching, and other models • Validate and confirm mapping basis user inputs • UI Screen to enter/modify DQ Rules • Create Data Dictionary to store all Business and Technical DQ Rules • Create Technical DQ Rules • Create Business DQ Rules • Execute Data Quality Checks • Dashboard to display DQ Score • Create Data Dictionary to store all Transformation Rules • Execute all standard data transformation rules (raw zone to consumption zone) • Capture customizations from the UI • Leverage Crayon and ecosystem sources for enriching entities in the data – merchants, locations, customers, etc. • Automatically identify entities and enrich using external corpus and entity resolution • Leverage GPT - led curation for brands, locations • Using a standard dictionary of metrics and features derivation logic, derive features and metrics in the Analytics ML zone • Store additional metrics in the metrics and feature stores for inferences • Pre - built analytical dashboards that provide insights on key aspects of the business • Self - serve dashboard creation for any custom insights • Pre - built ML models available for use - cases such as customer - lifetime value, card upgrade/x - sell, attrition prediction, destination prediction etc. • Models exposed as Data APIs for integrations, also available as flat - files for downloads

Launching Data Studio: Under the hood

| Data Studio components

| Data Studio components