ARTIFICIAL INTELLIGENCE SALES PITCH FRAMEWORK Agenda Introduction Approaches

ARTIFICIAL INTELLIGENCE SALES PITCH - FRAMEWORK

Agenda • Introduction • Approaches • Algorithm • Scope of AI • Technology • Thumb rules • Use cases

Introduction Artificial Intelligence: Machines/Applications performing "cognitive functions". Cognitive Functions: • • Reasoning Memory Attention Language Data: • Transaction data(predictive model) • Language data (NLP) • Image data(Image ID Model)

Approaches 1. Machine Learning 2. Deep Learning • Subset of AI and more advanced stage of Machine Learning • Self learning algorithms that detect patterns and anomalies • Don't need massive computational power to do this • This purely deals with big data • Most companies you’re selling to won’t be here because they lack the sheer amount of data

Algorithm/Model building stages Pipeline-Architecture 1. Storage 2. Ingestion 4. Train Algorithm 3. Pre-Process 5. Evaluate Algorithm/Model Machine Learning Deep Learning 6. Test 7. Deploy

Scope of AI/ML in Market-Sales Finance and Banking • Credit scoring • Fraud detection • Risk analysis • Trading exchange forecasting Retail and E-commerce • Demand forecasting • Price optimization • Recommendations • Fraud detection • Customer segmentation Marketing and Sales • Market and customer segmentation • Price optimization • Churn rate analysis • Customer lifetime value prediction • Upsell opportunity analysis • Sentiment analysis in social networks Travel and Booking • Demand forecasting • Price optimization • Price forecasting (for dynamically changing prices Healthcare and life Sciences • Increase in diagnostic accuracy • Identifying at-risk patients • Insurance product cost optimization Other • Object recognition (photo and video) • Content recommendations (movies, music articles and news) • And more

Top Players Technologies DOCUMENTS DATA PROTOCOL JSON, XML PDFS, TEXT FILES, UNSTRUCTURED FILES WEBSERVICE Storage Hadoop Lake Programming Tools Python, R, Scala No. SQL/Document Databases Cassandra, Mongo. DB Deployment Spark Cluster Open Source Frameworks

Thumb Rules AI…isn’t something you just install and walk away !!! 1 2 3 4 5 Know Your Challenge Make AI Reliable Sell Vision and Value Gain credibility for the long road ahead One thing that doesn't change AI is really just math, models. Revert very quickly to the pain points that technology was meant to solve AI is different from other solutions because it’s not a single piece of technology — it’s a composition. AI isn’t something you just install and walk away from. It takes a commitment, as the technology’s evolution is dependent on the customer’s usage. Of course, there’s one thing that doesn’t change when selling AI: taking care of the customer and being a trusted adviser. So your challenge and solution is to demystify this concept.

Solution: Challenge To be able to meet consumers’ needs, appropriate forecasting models are vital Understanding demand the ability to accurately predict dependency for efficient manufacturers, suppliers, and Retailers. Demand anomalies , Seasonal fluctuations. Domain: Retail, Travel &Booking Implemented a Predictive/Forecast Time series model using Deep Neural networks. Benefits/Inferences: • Assisting business decision makers with accurate insights • Other departments like Inventory mgmt. has accurate maintenance of stock. Reusable: • Demand Forecast • Spend Forecast

Solution: Challenge Identifying and extracting specific data from documents and placing them in relational/structured data base Additional Applications • Speech Recognition • Optical Character reading modelling • Time Series modelling Implemented a model using Natural Language processing with Bi. Directional LSTM Neural Network. Benefits: • Making extraction easy and converting it into relational data • Easy to establish a relation between drug and disease Reusable: • Speech Recognition • Optical Character reading(OCR) • Time Series modelling

Thank You