How to exceed customer expectations by delivering complex cognitive RPA projects.

Transcript

1. HOW TO EXCEED CUSTOMER EXPECTATIONS BY DELIVERING COMPLEX COGNITIVE RPA PROJECTS A N D R E I R OZ H N E U A true story of one project
2. Grodno, 2019 Agenda Grodno, 2019 1 EPAM IA PRACTICE 6 2 3 4 5 INTRODUCTION TO BUSINESS PROCESS FLOW CUSTOMER'S EXPECTATIONS RPA + ML APPROACH RESULTS AND STATISTICS INTELLIGENT AUTOMATION EFFICIENCY
3. Grodno, 2019 EPAM Intelligent Automation & RPA KEY FACTS DOMAINS Certified Developers and ML engineers 10+ 700+ Years of BPM/Robotics & Cognitive experience Global Services Locations 10+ PARTNERSHIPS & SOLUTIONS Robotic Process Automation (RPA) Business Process (BP) optimization Document Processing (incl. OCR) Cognitive Analytics, ML Virtual Assistant (Chatbot) C O N S U L T I N G E N G I N E E R I N G FINANCIAL SERVICES RETAIL AND DISTRIBUTION INSURANCE HUMAN RESOURCES TRAVEL AND HOSPITALITY ENERGY SOFTWARE AND HI-TECH HEALTHCARE
4. Grodno, 2019 Intelligent Automation (IA) Delivery Centers & Headcount CA US MX BY UA PL KZ RU CN IN AU 700+ INTELLIGENT AUTOMATION PROFESSIONALS 0 100 200 300 Belarus Ukraine India Poland USA China Mexico Canada Russia Kazakhstan Australia
5. Grodno, 2019 Introduction to Business Process Flow Grodno, 2019 ORDER DETAILS: • Customer ID • Shipment ID • Delivery Info • Pickup Info • Shipper Details • Receiver Details • Ship Instructions • Other fields Transportation Order
6. Grodno, 2019 Customer’s Expectations Grodno, 2019 MANUAL BUSINESS FLOW B E F O R E A F T E R AUTOMATED BUSINESS FLOW 00:40:00 00:10:00
7. Grodno, 2019 Robotics and Cognitive Hand work Robotics “aka” RPA i.e. entering data from one application into another Head work Cognitive Automation i.e. issue categorization, routing, resolution
8. Grodno, 2019 RPA + ML Approach • Email utils • Documents conversion • OCR • Manual Task customization • API call utils RPA • ML Model creation • ML Model training • Output fields format optimization ML Email Processing R P A Documents Conversion R P A OCR R P A Machine Learning M L Human Task M T Order Creation API Call R P A
9. Grodno, 2019 • Emails: Microsoft Exchange Web Services • File conversion: LibreOffice • File conversion: Apache PDF Box • OCR: ABBYY FineReader Engine Technology Stack • Internal system API
10. Grodno, 2019 Machine Learning Lifecycle Grodno, 2019
11. Grodno, 2019 AutoML SDK Grodno, 2019
12. Grodno, 2019 Manual Task Customization Grodno, 2019
13. Grodno, 2019 Machine Learning Results Grodno, 2019 Automaton – percentage of automatically extracted objects by Machine Accuracy – precision of extraction, calculated as the ratio of correctly extracted objects to all the extracted objects 200 DOCS TEST SET UAT EXECUTION
14. Grodno, 2019 Intelligent Automation Efficiency Order processing time from receiving an email with an order document till API Order submission Manual Task completing time from accepting a Manual Task by a worker till submission form 00:10:00 00:03:00 E X P E C T E D A C T U A L
15. THANK YOU FOR LISTENING! IT’S TIME FOR YOUR QUESTIONS