Intelligent Automation

Transcription

1 Intelligent Automation A component of a business transformation journey December, 2018 The better the question. The better the answer. The better the world works. Page 0

2 Agenda 1 Introduction and Definitions 2 Overview and Demonstration 3 Functional use cases & client case studies 4 Implementation Considerations Page 1

3 1 Automation Introduction Page 2

4 Megatrends that have transformed the workforce Offshore labor arbitrage and outsourcing drove a new round of cost savings by lowering the human costs of performing the associated services. Intelligent Automation - The next wave of cost savings is gathering pace, focused on replacing selected manpower with technology, evolving from desktop automation to RPA to cognitive automation. ERP and shared services - fuelled the emergence and growth of centralized finance and accounting, HR, procurement, and other business functions All components above have a place in the future state architecture of most organizations. Many companies start with the question, What can we eliminate or optimize through automation and then determine how Shared Services, Outsourcing and the extension of existing systems play a role around the remaining activities. Page 3

5 Types of Intelligent Automation today Execution Cognition Need Solution Assist Replace Assist Replace Assist staff with process execution at their workstations Remove staff from rules-based process execution Assist staff with decision making by filtering and providing quality data Automated decision making based on knowledge acquired from past experiences Desktop automation Unattended Machine Learning Scripting of individual tasks Runs on user s desktop Increases efficiency of workers Consolidates information and provides consistent experience Streamlines work and optimizes processes Large scale unattended processing Must respond to fluctuation in system response, unknown events, unanticipated business scenarios without interruption Considers security, scheduling, audit, exception management Secure, centralized collection of management information, audit records, process logs Requires only a few human to support many robotic activities Aids or replaces subjective decision-making based on large data samples Interprets contextual information and provides consistent reasoning Helps streamline of processes and route inquiries Applies human-like reasoning in large volumes (e.g., transaction monitoring, fraud identification, call filtering) Page 4

6 Types of Intelligent Automation today Robotic Desktop Automation (RDA) Need Solution Execution Cognition Helps people work faster and more efficiently Assist Replace Assist Replace Assist staff with process execution at their workstations Remove staff from process execution in the database Assist staff with decision making by filtering and providing quality data Remove staff with cognitive learning in the data center Desktop automation Unattended Machine Learning Scripting of individual tasks Runs on user s desktop Increases efficiency of workers Consolidates information and provides consistent experience Streamlines work and optimizes processes Overview Use Cases Large scale unattended processing Applications Must respond to fluctuation in system response, unknown events, unanticipated business scenarios without interruption Technologies Considers security, scheduling, audit, exception management Secure, centralized collection of management information, audit records, process logs Requires only a few human to support many robotic solutions Front office activities (shares desktop) Assisted sign-on Customer 360 view Activity logging Aids or replaces subjective decision-making based on large data samples Call centers, support agents, help desks OpenSpan/Pega Robotics, UiPath Interprets contextual information and provides consistent reasoning Helps streamline of processes and route inquiries Applies human-like reasoning in large volumes, (e.g., transaction monitoring, fraud identification, call filtering) Page 5

7 Types of Intelligent Automation today Robotic Process Automation (RPA) Covered Today Page 6 Need Solution Execution Repeatable Cognition rule-based automation Assist Replace Assist Replace Assist staff with process execution at their workstations Remove staff from rules-based process execution Assist staff with decision making by filtering and providing quality data Remove staff with cognitive learning in the data center Desktop automation Unattended Machine Learning Scripting of individual tasks Runs on each agent s desktop Increases efficiency of workers Consolidates information and provides consistent customer experience Streamlines work and optimizes processes Large scale unattended processing Must respond to fluctuation in system response, unknown events, unanticipated business scenarios without interruption Considers security, scheduling, audit, exception management Secure, centralized collection of management information, audit records, process logs Requires only a few human to support many robotic solutions Overview Use Cases Application Technology Back office activities Applicable to any process BluePrism, Automation Anywhere, Redwood, UiPath Aids or replaces subjective decision-making based on large data samples Interprets contextual information and provides consistent reasoning Helps streamline of processes and route inquiries Applies human-like reasoning in large volumes, (e.g., transaction monitoring, fraud identification, call filtering)

8 Types of Intelligent Automation today Intelligent Process Automation Overview Need Use Cases Applications Solution Technologies Execution Supports human interactions and decision making Cognition Assist Replace Assist Replace Assist staff with process execution at their workstations Remove staff from process execution in the database Natural language processing Automated support agents chat bots Agent assistance Automated concierge Medical transcription Assist staff with decision making by filtering and providing quality data Automated decision making based on knowledge acquired from past experiences Desktop automation Unattended Machine Learning Scripting of individual tasks Runs on each agent s desktop Increases efficiency of workers Consolidates information and provides consistent customer experience Service desks, translation IPSoft Amelia, Alexa, ABBYY Streamlines work and optimizes processes Large scale unattended processing Must respond to fluctuation in system response, unknown events, unanticipated business scenarios without interruption Considers security, scheduling, audit, exception management Secure, centralized collection of management information, audit records, process logs Requires only a few human to support many robotic solutions Aids or replaces subjective decision-making based on large data samples Interprets contextual information and provides consistent reasoning Helps streamline of processes and route inquiries Applies human-like reasoning in large volumes, (e.g., transaction monitoring, fraud identification, call filtering) Page 7

9 Types of Intelligent Automation today Overview Need Use Cases Execution Mass intake of knowledge allows for nextstep action Cognition Assist Replace Assist Replace Assist staff with process execution at their workstations Remove staff from process execution in the database Solution Applications Healthcare, consumer, technology provides consistent customer experience records, process logs Technologies IBM Watson, Cognitive Scale, [24]7 Assist staff with decision making by filtering and providing quality data Automated decision making based on knowledge acquired from past experiences Desktop automation Unattended Machine Learning Scripting of individual tasks Runs on each agent s desktop Increases efficiency of workers Consolidates information and Streamlines work and optimizes processes Cognitive Computing Bulk data analysis Medical diagnosis Customer analysis and recommendations Advertising Large analysis scale unattended processing Predictive analysis Technology planning (BCP, outages, etc.) Must respond to fluctuation in system response, unknown events, unanticipated business scenarios without interruption Considers security, scheduling, audit, exception management Secure, centralized collection of management information, audit Requires only a few human to support many robotic solutions Aids or replaces subjective decision-making based on large data samples Interprets contextual information and provides consistent reasoning Helps streamline of processes and route inquiries Applies human-like reasoning in large volumes, (e.g., transaction monitoring, fraud identification, call filtering) Page 8

10 2 Overview and Demonstration Page 9

11 RPA definition RPA as innovative solution for an automated execution of business processes What is RPA? RPA simulates an employee. The software robot has access to diverse applications with an ID or a password. The robot can gather information or change data. Consequently, business and administrative processes can be fully automated. RPA is integrated in an existing IT infrastructure. Robots deliver repetitive, deterministic, high-volume tasks efficiently, quickly, and consistently. People build relationships, provide subjective judgement, deliver lowfrequency tasks, and manage change and improvement. RPA is software. RPA is a computer software that runs repetitive, rule-based processes. The software is trained based on functional specifications and can be adjusted at any time. The RPA journey Page 10

12 Examples of robotics Bank Statement Reconciliation Digital enablement Recon Claims 1:42 Video Demo Rapidly performing repetitive tasks otherwise done by humans to reduce cost, accelerate timing, improve reliability and reduce risk 2:19 video demo Adding digital/mobile to applications, such as customer preference/profile, sales or service transactions Video Demos. Page 11

13 Benefits as a result of RPA Clear, traceable RPA benefits are reducing operating expenses, empowering the workforce and enabling speed to market Low risk Non-invasive technology Overlaid on existing systems and integrated with existing data minimizing disruption to existing IT strategy and architecture. Automation technology can begin with simple rules based tasks and scale to more sophisticated algorithms and machine-learning functions as the organization matures. Accuracy The right result, decision or calculation the first time Right shoring Geographical independence reduces need to offshore jobs while still delivering cost savings Consistency Identical processes and tasks, eliminating output variations Productivity Freed up human resources for higher value-added tasks. Cost savings Ranging from 20-60% of baseline FTE cost Cross-system Across systems since it works through the user interface layer Reliability No sick days, services are provided 365 days a year Audit trail Fully maintained logs essential for compliance Retention Shifts towards more stimulating tasks Scalability Instant ramp up and down to match demand peaks and troughs ROI Typical RPA projects include multiple 6-12 week deployments but the program typically returns an ROI < 1 year Page 12

14 Robotics Process Automation is often confused with traditional automation test automation and BPM, though the features are different Traditional Automation Business Process Management (BPM) Test Automation Robotic Process Automation (RPA) Functionality Automate steps, rules and functionality in a particular application Manages end to end business process through workflow Used to execute functional or load/performance test scripts Replicate human behavior and execute non judgmental sequence of activities across applications Applicability Across all types of processes for a particular activity Across all types of processes Execution of application specific scripts in a nonproduction landscape Rules based, non judgmental processes Technology Custom developed for a specific use case and technology, involves specific technical knowledge Technically integrated (APIs, interfaces) with the other business applications Can be coded to technically integrate into backend, at the data layer or GUI level Technology agnostic and configurable by more technical business users Impact Faster processing, reduced error rate Better monitoring, stronger control Good for high volume testing Significant savings in FTEs, faster processing, reduced error rate Examples Excel macro, startup scripts Pega, IBM, Activiti QTP, LoadRunner, Selenium Blue Prism, Automation Anywhere Page 13

15 Page 14 3 & Functional Use Cases Client Case Studies

16 RPA is usually the best starting point because it impacts the highest percentage of processes and is easier to implement and maintain Here is a natural progression from rules-based to more cognitive approaches where systems learn through experience, and can improve their performance beyond their programming. Chatbots E.g., Kore, Conversable Artificial Intelligence (AI) E.g., Watson, Holmes Robotic Process Automation (RPA) E.g., Automation Anywhere, Blue Prism, UiPath Cognitive RPA (including machine learning, natural language processing) E.g., Azure, Arago, Work Fusion 15% of process activities 15% of process activities 10% of process activities 60% of the enterprise's process activities Page 15

17 Opportunities for a virtual workforce span business functions Process characteristics to consider for RPA High, repetitive transaction volume The application scope is broad penetrating finance and accounting, treasury, tax, human resources, IT and supply chain High manual data entry Multiple systems to perform a task Multiple tasks to perform a process Finance and accounting Sales order Order to cash Collection Procure to pay Incentive claim Record to report Vendor setup Trend tracking Treasury Fx management Liquidity management Cash management Capital strategy Bank reconciliations Global economics Tax Scenario planning Update and maintain data Estimate, calculate, prepare tax provision Update and review effective tax rate Tax SOX compliance Identify and maintain tax payments E-filing Data entry and validation User interface navigation Automated formatting Copy and paste operations Login and logout of applications and ing Activities typically performed by RPA Human resources Payroll Benefits administration Pay slip management Time and attendance management Recruiting process Onboarding Education and training Compliance reporting IT Data synchronization Folder synchronizing, deleting and managing System installation Data transfer, download, upload or backup Server and app monitoring File management processing Batch processing Supply chain Work order management Demand and supply planning Quote, invoice and contract management Returns processing Freight management Page 16

18 Automation "hot spots" for Finance Financial Planning & Analysis (F,P&A) Automating the pre-population of forecasts using historical and market data Loading pre-populated balances into the planning system Creating variance reports to pre-population and to actuals 10 1 Accounts Receivable processing Credit approvals & customer master file maintenance Order processing A/R cash receipts processing & sending late notices via Regulatory reporting Data capture and cleansing to support automated generation of regulatory reports Pre-populating complex annual reporting 9 2 Accounts payable processing Vendor set up and maintenance Automating the workflow processes and approvals Data entry and payments preparation Automating processing of payments and bulk payment files for journal entries to sub system Financial review prep 8 Automating the preparation of management review slide decks by collecting data from multiple finance systems and reports Intercompany reconciliation 7 Automated checking and reconciliation of intercompany balances Basic research and reporting for exceptions Creating exception file and report for finance review and approval Account reconciliations 6 Automation hot spots Automating download of subaccount balances into preapproved format Upload detailed transaction data from various sub systems Perform data validation and basic research for exceptions Creating balancing journal entries to handle discrepancies 5 3 Operational finance and accounting Automating pricing reviews based on customer contracts and pre-approved price lists Calculation and processing of rebates Downloading of detailed monthly sales data and calculation of commissions Creating files and s to gain approvals 4 Posting to detailed sub systems and General Ledger Standard Journal entries Creation of standard monthly journal entries using prepopulated templates provided by different business users Performing validation analytics Posting to ERP Bank reconciliations Automating the download of bank statements for individual accounts Creating text files and storing in appropriate folders Reconciliation of balance and transactions to core finance sub systems Creating balancing journal entries to handle discrepancies Page 17 Finance functions face regular peaks in demand that could be supported through the use of robotic assistants. Automation of a range of core finance Automation activities of has the the Intelligent potential Enterprise to improve quality and allow great focus on analysis.

19 Automation can be used in project activity as well Have seen automation used in data migration/conversion, cutover execution, end-to-end process testing Robotic conversion advantages over a pure ETL technique Benefits We offer a unique perspective for converting legacy data using the power of robotics. Our approach allows you to reach your objectives faster, better and with less risk than other alternatives. Our team understands how to handle complex situations and are ready to serve as a trusted advisor throughout this journey Reduce data attributes for conversion as result of functional screen-based data sourcing and load Robots can be configured for data conversion much faster than a traditional technical ETL tool Uses existing and trusted user interface/functionality of the source system for data gathering and target platform for data loading Alleviates risk of missing nuances in the source system when gathering and transforming data Functional data conversion approach greatly reduces need for deep IT knowledge of source systems Sits on top of existing source and target systems and hence is less disruptive to current environment Faster and Faster and less intrusive less intrusive to MetLife Improved data quality Less risky; inherently Inherently simpler Page 18

20 How to get started with RPA Gate 1 Start 4-6 weeks A) Proof-of-Value Identify and prioritize automation opportunities with high level benefits Time-boxed POV focused on sample process/activity and connecting to a specific set of systems Built using evaluation SW license and delivered using dummy data in test environments Triggered manually where required to promote clear entry and exit points Stubbed where required to demonstrate functional and technical concepts Show-cased with key stakeholders to secure commitment for an prod pilot Gate weeks B) Production Pilot Builds on the outputs of the POV, improving outputs by removing stubs and adding sophistication and robustness Security protocols refined and implemented, and basic non-functional requirements agreed with IT Additional sophistication added into exception handling procedures IT & Business acceptance testing methods formalised and adopted in line with SDLC / governance SW procured and solution migrated to production environment with live trial audience using real data Scaled production architecture for applications / Infrastructure agreed Benefit case enhanced and detailed design of production environment commenced Gate 3 8+ weeks C) Enterprise Scale Scaled production architecture for applications / Infrastructure agreed Final prioritisation of global roll-out, release plan and benefits case agreed Multi LOB/Process solutions migrated to production environment Architecture defined and established in production SLAs and OLAs are clearly defined and agreed and solution operates within these boundaries BAU Operating model implemented Run & Maintain organizations fully operational Additional functionality is developed iteratively on development environments and, when ready, is migrated to production environment Page 19

21 4 Implementation Considerations Page 20

22 Risks and related control activities Risk domain Policy and governance Risk description A lack of robotics governance can lead to ineffective and inefficient process automation and an inability to support and meet business requirements. Illustrative controls Policies and procedures change management, access control, segregation of duties, operations, issue management, RPA center of excellence (COE) Ongoing monitoring performance, control processing and quality assurance Governance, risk and control risk and control requirements defined in RPA strategy and deployment, i.e., approval of new robots, approval of robot ID, development controls for change management process and access, user acceptance testing, migrate to production approvals Logical user access (GCC) Robotics access management is ineffectively managed, leading to the compromise of systems, applications and their associated data. Security privileged access to provision, de-provision and modify robot IDs is limited to COE; documentation is maintained Humans interacting in the same process do not have access to create robots or change robot processing (SOD) Robot permissions and profiles are restricted; audit logs are maintained of each robot user ID System change management (GCC) Robotics implementations are not appropriately designed and tested, leading to requirements not being met or a negative impact on production systems resulting in a negative impact on the business and financial losses. RPA development and change management key life cycle controls, authorization, testing, approval, restriction to change in production to COE members; business users can perform approvals Security privileged access to make robot changes, host system profile protection, restriction of use Humans interacting in the same process do not have access to make changes to robot instructions or tasks Page 21

23 Risks and related control activities Risk domain Risk description Illustrative controls Timely system outage/issue detection Automation problems are not timely identified and managed, leading to a delay in their resolution and resulting in a negative impact to business processes. Interface and system error reports are generated and reviewed periodically to verify robots are running as planned and gathering the planned data through interfaces Human review of issue and error reports and identifying next steps Security privileged access to correct system issues limited to COE and documented as incidents Vendor/ third-party management Risks are not effectively mitigated for robotics vendor relationship and outsourced services, leading to financial and reputational exposure. SOC report reviews, right to audit clauses, appropriate SLAs, defined maintenance contracts, limited vendor access with monitoring Page 22

24 Lessons learned from our journey: Ten common RPA issues 1 Not considering RPA as business-led, can t be IT driven 2 Treating robotics as a series of automations vs. an end-to-end change program 3 Not having an RPA business case or postponing until after proof-of-concepts or pilots 4 Underestimating what is needed to execute processes once they have been automated 5 Applying Targeting RPA at the wrong processes 6 traditional delivery methodologies (not agile) 7 Automating too much of an as-is process and not optimizing for RPA 8 Assuming RPA is all that is needed to achieve ROI 9 Forgetting about technology and the IT organization 10 Assuming capabilities needed for a pilot are sufficient enough for enterprise automation Page 23

25 Intelligent Automation For more information, please contact: Jon Smith EY Advisory Page 24