·
Cursos Gerais ·
Outros
Envie sua pergunta para a IA e receba a resposta na hora
Texto de pré-visualização
Robotic Process Automation A Lean Approach to RPA Carina Maria Goncalves Martins carinamartinstecnicoulisboapt Instituto Superior Tecnico Lisboa Portugal November 2018 Abstract Automation is not a new concept in organisations as a way to improve their processes However Robotic Process Automation is an emerging form that automates processes with software which the industry calls robots These robots perform repetitive and lowcomplexity tasks previously performed by humans in front of a computer perhaps the most commonly used feature in a company nowadays Using the Design Science Research methodology to build up this thesis it argues that RPA is being used idly compared to what it could be using process improvement techniques before applying the automation itself Thus this thesis proposes a new approach to RPA using techniques of Lean This thesis assesses two market leaders on RPA and suggests a framework of activities for organisations that are investing in RPA and that want to take advantage of the capabilities that this technology currently offers The majority of the proposal demonstration was done in a Portuguese private bank in three processes It had an overall positive evaluation in the field or the simulations depending on the different cases Comparing RPA and Lean RPA projects in the amount of resources time FTE needed to carry out business processes the latter approach presented values significantly lower and consequently satisfactory Keywords Robotic Process Automation Lean Management Lean RPA Process Automation Business Processes Continuous Improvement 1 Introduction A fourth wave of technological advancement is be ing witnessed a new digital industrial technology known as Industry 40 Technologies are already positively impacting productivity and employment on large enterprises 17 That is the path this paper follows In the enter prise world theres a ceaseless need of doing more with fewer resources as possible This objective cre ates a high demand for continuous improvement in the business processes of organisations using ra tionalisation and optimisation of the resources 12 Accordingly this is the beginning point for Robotic Process Automation RPA The term automation will be employed as the use of scientific and technological principles to replace businessrelated tasks previously done by humans The term process will be used to define a collection of tasks taken in a specific arrangement to achieve a determined output 11 Robotic Process Automation RPA is an emerging form of process automation whereas one or more software robots perform the exact same procedure as a human would do The Robotic part in the term only emphasises the idea of a machine doing utilities instead of a human worker as in its not a real physical robot Robot is the concept used in the RPA industry and it will be the concept used throughout this paper to define the software robot By using robots there is a reduced risk for cross organisations errors as well as higher availability since robots can work 24 hours a day 7 days a week and will never get sick or need vacations bringing faster results Currently the tool offers basic digi tisation and enhanced digitisation not being able yet to perform cognitive decision management 12 Lean The origins of Lean come from Japanese manufac turers in the 1950s and consists of the following principles the identification of the value the elimi nation of the identified waste activities the genera tion of flow pulling work instead of pushing it and continuous improvement in the organisation while reaching a costvalue equilibrium 13 To progress the reading of this paper its neces sary to define certain used concepts Value is some thing the customer is willing to pay for Flow is 1 used as a concept to describe how work progresses in the system 13 Waste on the other hand should be considered as everything that does not add value Waste can be found in any activity of the flow 13 Research Methodology Paper Outline The research methodology that is used throughout this work is the Design Science Research Method ology DSRM 5 DSRM aims to solve identified organisational problems by designing and evaluat ing IT artefacts It consists of six activities as seen in the figure 1 Figure 1 Detailed DSRM for this work Adapted from 14 This paper begins with a brief Introduction sec tion 1 about RPA and Lean It advances to the discovered Research Problem section 2 Section 3 is the related work for this paper In section 4 the Research Proposal and the Objectives are de tailed and its application is described in section 5 Demonstration Follows with an Evaluation of the results section 6 and concludes in section 7 2 Research Problem Nowadays RPA is being used as an almost instan taneous automation tool because it delivers quick results and doesnt involve changing the already ex isting underlying computer systems inside organisa tions 7 RPA developers dont usually care about optimisation 8 since that as long as it mimics the work that has been successfully done by human workers the project is sold But what if RPA tech nology was more present inside an organisations processes Two of the questions raised in this paper are how can RPA take advantage of a continuous process im provement approach and what is its impact To summarise what has been stated the main problem found is that RPA is being taken as a quick fix existing a lack of detailed processes analysis undervaluing its value to organisations when com bined with a process improvement approach such as lean 3 Related Work This section presents background information about RPA and Lean tools 31 RPA Life Cycle The currently RPA development model widely used is expressed in Fig 2 obtained from Anagnoste 2018 1 and Blue Prisms Guide 16 The main tenance phase is not represented Figure 2 Currently most used RPA development model from 1 First theres a gathering of the processes suitable for automation and then it proceeds with choosing one A specific RPA report from EY 4 enhances three key characteristics for a strong RPA candi date The actions need to be consistent the pro cess should be template driven and rulesbased A technical report from Symphony Ventures 2 utters an additional characteristic must be rawbased ie does not require human judgement element Next the goal of Process Reengineering is also to provide a highlevel analysis of the process solution the automation efficiency and estimation of effort 16 Following the project approval comes a detailed study where the RPA developer gets in detail with chosen process After having a functional documen tation and the project plan defined it is time for building the robot and test it There is no average time for the total time of an RPA project since it depends on the selected processes and its business complexity 32 RPA Tools Available in the Market For the past couple of years only Forrester Research 9 is doing research on RPA tools on the market and publicly communicate the results According to Forrester Researchs RPA Report from 2017 8 and from 2018 9 the market lead ers are Automation Anywhere Blue Prism and UiPath although currently there exists more than 15 tools 33 Lean 8 Wastes Lean tools are known for actively trying to reduce the following eight wastes 19 10 Transporting Excess Motion Waiting Overproduction Unneces sary Inventory Over Processing Defects and Skills 34 Lean Kaizen Kaizen is a Japanese word that translates to contin uous improvement Kaizen consists of small steps for each improvement whilst the traditional man ner mostly improves in substantial form causing high costs and affecting most people in the organi sation Regarding potential and benefits kaizen in vokes workers to participate in the upgrade from the 2 beginning unlike traditional improvement where the idea to change stays with the organisation top management until it goes live 11 35 Lean SMART Goals The appliance of lean tools to a system must have a reason goals must support that reason The term SMART Goals has been on trend for a long time and literature is abundant about its definition and usage Conzemius et al 2009 3 define SMART goals to be Specific Measurable Attainable Relevant and Timespecific 36 Lean Value Stream Management Value Stream Management is a process for planning and linking lean initiatives through data capture and analysis and consists of a guideline that a company must follow to achieve lean processes Tapping et al 2002 18 asserts that the guideline is suitable for office processes that is processes that dont pro duce a physical product but instead provide value to the final customer through a value stream com posed by work units 4 Research Proposal In this section the proposal objectives and the pro posal description are defined 41 Proposal Objectives This section is connected to the Definition of the Objectives for a Solution activity of the Design Science Research Methodology The primary objective this work aims to achieve is to understand how much more effective and effi cient Lean RPA1 is comparing to a standard RPA project To help solve that statement the following sub objectives must be achieved 1 RPA Tools Assessment In order to proceed to the next subobjectives its necessary to choose one tool to work with The two possible tools to work on are two of the current market lead ers BP and UiPath 2 Framework for Lean RPA This work defends that analysing and changing processes should be taken into account before the development of any RPA project This being said a frame work that joins both lean and RPA lifecycles must be created 3 Collection of RPALean RPA Results Gather results of applying both RPA and Lean RPA either in simulation or realfield in order to compare both approaches in terms of process ing time 1Lean RPA is now the created term for RPA when com bined with a lean approach Accomplishing these three subobjectives will provide enough results to answer the primary ob jective 42 Proposal Description This section is connected to the Design Devel opment activity of the Design Science Research Methodology The proposal model is represented in Figure 3 This model helps to visualise how the solution un rolls following the previously written three sub objectives The trigger Desire to Change is represented only to enhance a beginning It exhibits the sense of change and innovation that is necessary for pro cesses improvement projects If this trigger is ne glected then the following information is meaning less since it is based on continuous improvement Figure 3 Proposal model 421 RPA Tools Assessment Criteria The criteria for assessment was based on the criteria that Le Clair et al 2018 8 proposed but also on ISOIEC 25010 characteristics 6 that were mapped to RPA Functional Suitability Functional Appropri ateness Thirdparty Integrations Functional Efficiency Time Behaviour Error proofing Mechanisms Usability Previously required programming skills Overall Usability and available Training courses Reliability Robustness Security Robots deployment Scheduling and Logs quality 3 Maintainability Reusability of modules Portability Performance of the program on different information systems IS and perfor mance of robots 422 Lean RPA Framework Improvement Cycle The Improvement Cycle is detailed on the Figure 4 Each one of the four steps has a set of activities below that must be completed before moving on to the next step Those activities should be done in the order theyre presented 1 Process Selection The first step has activities are very centred on lean principles whilst the decision point starts tak ing RPA into account Lean suggests that defining a value stream VS is the initial point for change Next processes that belong to the VS should be sorted and one must be embraced to further mapping When mapping processes its fundamental to be simple yet detailed Everyone involved in the pro cess should be able to understand and identify waste tasks with at least one the eight wastes of Lean If the chosen process requires mainly cognitive actions even after identifying waste to be removed then its not suited for RPA and the first step should be repeated Otherwise its viable to continue to step 2 2 Process Study This step starts with the RPA Analysis activity Its necessary to assert what RPA can or cannot do in the process based on the current functionalities it offers Once that is finished its time for defining kaizen plans Kaizen is the correct choice here since it incrementally changes the process to consolidate with the RPA solution The next activity is mapping the future state Mapping the future state will help developing the robot implementing change to the process flow and having the new process documented Furthermore after having some documented processes improved by the improvement cycle it might become effort less to find similarities between processes and start standardising it Actions that should be redirected to the robot must be stated in the lowlevel map Before the implementation its also indispensable to define SMART goals for the project Without goals one could not understand if it was successful or not The goals should also be documented to be revisited later in step 4 3 RPA Project Development Its the step where the process flow changes should occur and the robot should be developed This step is on constant repetition so the process doesnt stop working another reason for kaizen plans After some iterations of the improvement cycle the developers might notice some similarities be tween processessubprocesses If thats the case this situation deserves an analysis to see if the used methodology can become a standard for the organ isation While implementing the robot standard opera tion procedures SOP and Simulations should be used to observe and test the development These activities avoid advancing to step 4 with errors and performance issues 4 Maintenance Review After the RPA Project Development its nec essary to review the goals For this one should compare the obtained results with the goals de fined in step 2 If the project achieves those goals the project goes live and a maintenance activity to that project shall be supported from now on If the project doesnt comply with the goals from step 2 its necessary to go back to step 2 and check what went wrong Finally the cycle ought to continue since lean proclaims for continuous improvements 423 Collection of RPALean RPA Results Once the improvement cycle has had enough pro cesses the results for each selected process should be gathered through simulation for RPA approach and realfield logs for Lean RPA approach 5 Demonstration This section connects to the Demonstration ac tivity of the DSRM 51 RPA Assessment Results The methodology used to gather information about the selected RPA tools was based on the analysis of the characteristics presented in technical reports by the vendors Forrester Wave reports 9 8 partici pation on official online learning courses communi cation with other professional RPA developers and more than one year of experience developing robots for professional and recreational purposes The results are stated below and Table 1 briefly summarises what has been assessed and the prices for each tool per year It is assessing UiPath without Orchestrator no scheduling and additional costs and without the additional cost of acquiring the development environment for more 3k Table 1 shows that its safe to assert that UiPath is more adequate for entrylevel Its cheaper and possesses an active online community and free train ing However being BP more robust its more adequate for processes dealing with audits frauds situations etc It takes longer for a beginner to 4 Figure 4 Lean RPA Framework Improvement Cycle Table 1 Analysis summary for the two assessed tools Parameters BP UP Functional Suitability 3rd Party Integration Time Behaviour ErrorProofing Mech Programming Skills Overall Usability Training Courses Robustness Scheduling Logs Quality Reusability Performance on Df IS Price per robot 18k 6k learn due to its complexity Its price is the triple of UiPaths price but that can be reasoned with the Scheduling feature which allows robots to work 247 for processes that require it 52 Lean RPA framework Improvement Cycle The demonstration for this subobjective was per formed inside a private Portuguese bank The infor mation gathering for the following subsections was done not only using indirect and direct observation but also reading documented processes and ques tioning employees about their routine at work RPA was already being used to tackle backoffice tasks In this demonstration processes will be modelled using BPMN due to its convenience on understand ability and on simulations 521 1 Process Step 1 Process Selection Choose Value Stream Firstly it was necessary to assess the value in the bank banking services For every bank service its necessary to open an account The ABC team is the responsible and seemed a good place to start The bank has one clear objective reduce as many FTE from ABC team as possible The ABC team deals with new accounts and any alterations to accounts that may occur Choose Process There were some chores stated as obnoxious and no one ever wanted to do it The most cited process was the DMIF Forms2 Fulfil ment This process consists of receiving the notifi cation that some customer updated hisher DMIF form and proceed to manually update every form in every account the customer is inserted Map Current State Identify Waste The pro cess is mapped in Figure 5 The activity in purple is the identified waste Searching for new DMIF alter ations can be considered as over production since sometimes there are no alterations for that day and time was just wasted in that activity Nonetheless the whole process is mechanical and requires zero cognitive actions This is directly mapped to the 8th waste of lean skill Once settled the project advanced to step 2 process study Step 2 Process Study RPA Analysis Define Kaizen Plans Search ing for new DMIF alterations is something RPA can easily do However the robot was not able to screenscrape the new answers from the form This 2DMIF Forms are forms that ask customers about their financial experiencie Its obligatory to be fulfilled for every account and always uptodate 5 Figure 5 Original DMIF Forms Fulfilment process was not a satisfactory solution since it couldnt re move any waste Knowing RPA current capabilities and limitations the IT department was contacted to know if there was a database with the answers It was affirmative but would take some time to give access to the robot The process could not stop working whilst being changed Thus the defined plan was to start with the implementation described previously This way workers would not waste time searching because the robot would send an email with the information if there were any alterations Whenever IT would give access to the database the rest of the project could continue Map Future State Detailed Study of the Pro cess The future state is mapped in Figure 6 The activity Search for new DMIF alterations is still considered waste since there might be no new alterations since the last check However it is now attributed to the robot which takes seconds to do Also if there are new DMIF alterations it instantly proceeds to do it Define SMART Goals The robot must be able to perform one instance of the process from the searching to one DMIF alteration in less than five minutes Its also intended that the DMIF updates are performed in every customers account within twentyfour hours of the original alteration from the customer Additionally after the final project goes live the ABC team can no longer interact with the normal flux of the process Once settled the project advanced to step 3 implementation Step 3 RPA Project Development Implement Kaizen Plans Develop Robot Cre ate SOP and Simulation Observe Performance First the ABC team started getting emails from the robot about any DMIF alterations With those emails ABC used the information to update the DMIF As soon as the IT department gave access to the database the rest of the project was developed within one week Step 4 Maintenance Review Figure 6 Finished DMIF Forms Fulfilment process with Lean RPA Review Goals The first and last goal were met as detailed later in next subsection For the second goal it was also met by merely running the robot every 24 hours Thus there was no need to regress to step 2 Go Live Maintenance The process went live This generates logs that are useful for further main tenance and to obtain results for evaluation The first process using the cycle was then seen as completed and more processes were on the queue for RPA 522 2 Process Step 1 Process Selection Choose Value Stream Choose Process Map Current State The DMIF Forms Fulfilment process reminded ABC team of another search they need to make every day unfailingly The bank possesses three types of passwords for accounts When a customer requests for missing password or wants to change one a request is gen erated on the banks intranet The ABC team is then responsible for searching new requests and re turning them to the customer Identify Waste The identified waste on the pro cess was the skill waste since searching for pass word requests is direct and mechanical This process requires cognitive actions to at tribute passwords however searching for passwords requests is direct and rulesbased Therefore this subprocess progresses to step 2 Step 2 Process Study RPA Analysis Define Kaizen Plans Since the robot was able to search effectively for the DMIF Forms Alterations in the last process this sub process is the same procedure but for another type of request Map Future State Detailed Study of the Pro cess Sending Passwords will now begin its flow with the robot The robot sends an email for ABC with the requests from the past twentyfour hours 6 and the team continues the original flow accord ingly Define SMART Goals This subprocess should take no longer than one minute Additionally every request should be completed in the following twenty four hours on weekdays Step 3 RPA Development Project Implement Kaizen Plans Develop Robot Cre ate SOP and Simulation Observe Performance This subprocess was developed straightforwardly since it only needed to run a query obtain the results build a table with the necessary informa tion and send it to ABC by email This can be come a standard methodology for upcoming similar processessubprocesses Step 4 Maintenance Review Review Goals The first goal was met as stated later in the next subsection The second goal was also met By sending an email to ABC with all necessary details its easier for the team to know how to carry on Go Live Maintenance The process went live This generates logs that are useful for further main tenance and to obtain results for evaluation The second process using the cycle this time sim pler but useful was completed 523 3 Process Step 1 Process Selection Choose Value Stream Choose Process Accord ing to the first process and bank intentions ABC team is still the chosen value stream The ABC team consists of 4 members and every working hour there must be at least 2 members validating docu ments This is because every operation related to accounts requires it Thus the next process should be one to relieve this burden One of the most prominent processes in the bank is the Accounts Opening This subprocess con sists of a Portuguese customer entering in one of the bank agencies and proceeding to the open an account with the commercial person After this the commercial person sends all documents back to ABC team to validate basic information on them and to open the account Map Current State Identify Waste The origi nal state is mapped abstractly highlevel in Figure 7 In the Figure 7 the waste is represented in pur ple in the Commercial lane Contact customer for missing documents and Receive missing docu ments are activities of over processing and defects waste The commercial person is obligated to re peat the action of digitalising documents and send ing them to ABC The orange activities are the skill type of waste Figure 7 Original Accounts Opening in the Agency process Despite the marked waste activities theres also inventory waste The bank keeps every document even whats not necessary to open an account or the customer stops sending the remaining documents being idle Although the commercial part requires cogni tive actions validating documents is straightfor ward since its only necessary to read the document and check if it conforms with the rules Therefore the project advanced to step 2 Step 2 Process Study RPA Analysis Define Kaizen Plans Applying straightforward RPA development to the process would entail ABCs verification for every account due to todays OCR low reliability This would not be practical nor reduce any waste However check ing documents and sending the process to conclu sion is accessible for RPA and reduces some working time on ABC team members Nonetheless the process flow needs modifica tions Map Future State Detailed Study of the Pro cess The future state is mapped in Figure 8 The most significant change is in the Commercial lane instead of the Robot The commercial checks with the customer for the necessary documents before starting signing up If the customer doesnt bring the mandatory docu ments then its informed by the commercial person on the missing documents and how to proceed with the signing up Afterwards the robot just needs to open the account by checking the documents and send the opening process to conclusion As seen the ABC team is now spared from this process Define SMART Goals This process needs to re duce its average time by half to spare one FTE Additionally after the final project goes live the ABC team can no longer interact with the normal flux of the process Once settled the project was meant to advance to step 3 However it didnt progress any further The Lean RPA results presented in the next section are obtained through simulation 7 Figure 8 Accounts Opening in the Agency process with Lean RPA 53 Collection of RPALean RPA Results To collect RPA results the only RPA processes were designed and simulated for twentytwo days one month This was also used to collect Lean RPA results from the 3 Process The Lean RPA results were collected on the field for one month Every time the robot would run a certain process or subprocess it generated a log with what was done and the processing times 531 DMIF Forms Fulfilment Results Table 2 shows the obtained results from the DMIF Forms Fulfilment process The written values are the number of minutes wasted on the process per weekday Keep in mind that the average of DMIF alterations is four but this is not a reliable value since the process is initially triggered by the cus tomer and therefore is not constant Table 2 DMIF Forms Fulfilment process time re sults in minutes per day without waiting time Team Original RPA Lean RPA ABC Avg 27 195 0 Min 15 0 0 Robot Avg 011 076 Min 008 008 532 Searching for Password Requests subprocess Results Table 3 shows the obtained results from the au tomated subprocess Searching for Password Re quests This table doesnt include a column Pro cess w RPA because investing in automation for a subprocess as simple as this one would not be re warding This was only automated because of the lean approach Table 3 Searching for Password Requests sub process time results in minutes per day Team Original Lean RPA ABC Avg 15 0 Robot Avg 1 533 Account Opening Results Table 4 shows the obtained results in the real field original and simulation RPA Lean RPA from the Account Opening process The written values are the minutes wasted on the process per every new account Minimum values are ignored for Lean RPA due to the absence of work repetition Table 4 Account Opening process time results in minutes per new account without waiting time Team Original RPA Lean RPA Com Avg 32 32 36 Min 275 275 ABC Avg 125 787 0 Min 725 5 Robot Avg 176 07 Min 176 6 Evaluation This chapter is connected to the Evaluation ac tivity of the Design Science Research Methodology It follows some of the evaluation criteria proposed by 15 such as Efficacy Generality and Validity from the Goal dimension and Utility Fit with Or ganisation Harnessing of Recent Technologies from the Consistency with organisation criterion from the Environment dimension 61 1 SubObjective RPA Tools Assess ment The bank had already bought a RPA license UiPath However the banks reasons correspond to the results of the assessment It was due to its price and overall appropriateness and usability There fore the assessment reaches the validity criteria as it matches with the bank choices 62 2 SubObjective Lean RPA Frame work Regarding generality it determines how general is the artefact Its necessary for the proposed frame work to have a broader range of possible processes to act on to The framework was validated in the bank and there were no barriers in any phase How ever its not possible to guarantee that the evalua tion is entirely positive on this criterion because it needs to be tested on more realities 8 Table 5 Average time difference in hours in one month for each process without waiting time Process Original Process Process w RPA Process w Lean RPA 1 DMIF Forms Fulfillment 99 715 028 2 Searching for Password Requests 55 037 3 Accounts Opening in Agency 3263 3053 2691 After applying the solution in the bank and get ting positive results from the teams involved the bank will continue using it for its processes im provement Consequently the framework passes the following evaluation criteria utility and fit with organisation Finally it arose from the necessity to approach a new technology RPA in a different way than the companies have approached Thus the framework harnesses recent technologies 63 Primary Objective To recall the primary objective for the proposal was to understand how much more effective and efficient Lean RPA is comparing to RPA Regarding the efficacy criterion which is the de gree to which the artefact produces its desired ef fect the table 5 supports the primary objective evaluation Table 5 represents the three processes presented in Demonstration section 5 on monthly aver age processing time difference The following ta ble doesnt take into account the waiting time each process has Its assumed one month equals to 22 working days that the new accounts average is 20 per day and the average of DMIF alterations is 4 per day The sum of all the time saved using Lean RPA instead of normal RPA equals to little over six working days of one FTE for one month 2739 Although only three processes were studied the bank naturally has way more manual and rulesbased processes more specifically the backoffice department Continuing the au tomation journey reducing FTE will be organic to the bank Its adequate to assert that in these cases Lean RPA was more efficient and more ef fective than RPA Efficient because the processes reduced their processing time and effective because changing processes taking into account RPA current capabilities will reduce any runtime mistakes 7 Conclusions Using Design Science Research Methodology this work proceeded to explain the problem found with the typical RPA implementations and objectives Being RPA a lightweight technology it is a quick au tomation that may fix some problems but its value is being undervaluing by organisations To try to prove it a primary objective has been defined to understand how much more effective and efficient Lean RPA projects were in compari son to RPA projects Primarily it was necessary to assess an RPA tool Blue Prism and UiPath were used and put to test It was concluded that although BP is more robust and reliable UiPath is the most correct choice for RPA starters due to its functional appropriateness and price Secondarily to obtain Lean RPA results creating a framework to follow its steps was essen tial The framework has incremental activities aim ing for the participation of everyone involved since the beginning of the project To culminate after applying the first two sub objectives the outcome was evaluated with the gathered results The framework was applied in one bank for three processes and the results were mostly positive Despite the organic time reduction that RPA projects already secure every process had a significant reduction on time comparing RPA and Lean RPA It was also successful concerning its util ity since the referred bank has continued following the activities Concerning the two asked questions in section 2 Research Problem How can RPA take ad vantage of a continuous process improvement ap proach and What is its impact Lean RPA takes advantage of a continuous improvement approach by automating and changing processes and sub processes that would not be approached by simply RPA projects since it doesnt fit the typical pro cess choice The studied processes were of easy and secure development and implementation whereas other complex processes would take way longer Due to its price fast development and effectiveness Lean RPA projects can change the way processes flow inside an entire an organisation with the right improvements and right thinking on the processes modelling and execution It does not leverage the overall cost by much since it uses basic ideas and takes advantage of internal teams help that already know the processes and workflows To conclude the DSRM the activity Communi cation is covered by the respective dissertation and this paper that communicates the results 9 71 Limitations This work was limited in terms that it should be tested more and on different realities Additionally the proposed Lean RPA framework is not a stand alone tool 72 Future Work The Lean RPA framework needs to be evaluated for its understandability for the users Future efforts are also required for an improved framework such as merging Lean with Six Sigma Acknowledgements Id like to thank my parents for the support Thank you to my supervisor To my friends thank you for never failing to make me laugh To Pedro thank you for always believing in me And finally thank you to everyone that I met through the development of this work References 1 S Anagnoste Setting Up a Robotic Process Automation Center of Excellence Manage ment Dynamics in the Knowledge Economy 62307322 2018 2 I Barkin Best Practices Guide Roadmap to Robotic Process Automation Technical re port Symphony Ventures 2017 3 A Conzemius J ONeill C Commodore and C Pulsfus The Power of SMART Goals Us ing Goals to Improve Student Learning Solu tion Tree Press 2009 4 B Diepeveen J Matcher and B Lewkow icz Robotic process automation Automa tions next frontier Technical report Ersnt Young 2016 5 Hevner March Park and Ram Design Sci ence in Information Systems Research MIS Quarterly 28175 2004 6 H W Jung S G Kim and C S Chung Mea suring software product quality A survey of ISOIEC 9126 IEEE Software 2158892 2004 7 M Lacity and L Willcocks What knowledge workers stand to gain from automation Har vard Business Review pages 17 2015 8 C Le Clair A Cullen and M King The For rester WaveTM Robotic Process Automation Q1 2017 Technical report Forrester 2017 9 C Le Clair G ODonnell W McKeonWhite and D Lynch The Forrester WaveTM Robotic Process Automation Q2 2018 Technical re port Forrester 2018 10 J Liker The Toyota Way McGrawHill Edu cation 2003 11 A Manos The Benefits of Kaizen and Kaizen Events Quality Progress 4024748 2007 12 A Manuel Robotic Process Automation Uma analise comparativa das solucoes actuais 2017 13 T Melton The benefits of lean manufacturing What lean thinking has to offer the process in dustries Chemical Engineering Research and Design 836 A662673 2005 14 K Peffers T Tuunanen M A Rothenberger and S Chatterjee A Design Science Re search Methodology for Information Systems Research Journal of Management Information Systems 2434577 2007 15 N Prat I ComynWattiau and J Akoka Artifact Evaluation in Information Systems Design Science Research A Holistic View PACIS 2014 Proceedings Paper 23116 2014 16 B Prism Lifecycle Orientation 2016 17 M Rußmann M Lorenz P Gerbert M Wald ner J Justus P Engel and M Harnisch In dustry 40 The Future of Productivity and Growth in Manufacturing Technical Report April BCG 2015 18 D Tapping and T Shuker VSM for the Lean Office Productivity Press 2002 19 J P Womack and D T Jones Lean Thinking Banish Waste and Create Wealth in your Or ganization volume 5 Simon Schuster 1995 10
Envie sua pergunta para a IA e receba a resposta na hora
Texto de pré-visualização
Robotic Process Automation A Lean Approach to RPA Carina Maria Goncalves Martins carinamartinstecnicoulisboapt Instituto Superior Tecnico Lisboa Portugal November 2018 Abstract Automation is not a new concept in organisations as a way to improve their processes However Robotic Process Automation is an emerging form that automates processes with software which the industry calls robots These robots perform repetitive and lowcomplexity tasks previously performed by humans in front of a computer perhaps the most commonly used feature in a company nowadays Using the Design Science Research methodology to build up this thesis it argues that RPA is being used idly compared to what it could be using process improvement techniques before applying the automation itself Thus this thesis proposes a new approach to RPA using techniques of Lean This thesis assesses two market leaders on RPA and suggests a framework of activities for organisations that are investing in RPA and that want to take advantage of the capabilities that this technology currently offers The majority of the proposal demonstration was done in a Portuguese private bank in three processes It had an overall positive evaluation in the field or the simulations depending on the different cases Comparing RPA and Lean RPA projects in the amount of resources time FTE needed to carry out business processes the latter approach presented values significantly lower and consequently satisfactory Keywords Robotic Process Automation Lean Management Lean RPA Process Automation Business Processes Continuous Improvement 1 Introduction A fourth wave of technological advancement is be ing witnessed a new digital industrial technology known as Industry 40 Technologies are already positively impacting productivity and employment on large enterprises 17 That is the path this paper follows In the enter prise world theres a ceaseless need of doing more with fewer resources as possible This objective cre ates a high demand for continuous improvement in the business processes of organisations using ra tionalisation and optimisation of the resources 12 Accordingly this is the beginning point for Robotic Process Automation RPA The term automation will be employed as the use of scientific and technological principles to replace businessrelated tasks previously done by humans The term process will be used to define a collection of tasks taken in a specific arrangement to achieve a determined output 11 Robotic Process Automation RPA is an emerging form of process automation whereas one or more software robots perform the exact same procedure as a human would do The Robotic part in the term only emphasises the idea of a machine doing utilities instead of a human worker as in its not a real physical robot Robot is the concept used in the RPA industry and it will be the concept used throughout this paper to define the software robot By using robots there is a reduced risk for cross organisations errors as well as higher availability since robots can work 24 hours a day 7 days a week and will never get sick or need vacations bringing faster results Currently the tool offers basic digi tisation and enhanced digitisation not being able yet to perform cognitive decision management 12 Lean The origins of Lean come from Japanese manufac turers in the 1950s and consists of the following principles the identification of the value the elimi nation of the identified waste activities the genera tion of flow pulling work instead of pushing it and continuous improvement in the organisation while reaching a costvalue equilibrium 13 To progress the reading of this paper its neces sary to define certain used concepts Value is some thing the customer is willing to pay for Flow is 1 used as a concept to describe how work progresses in the system 13 Waste on the other hand should be considered as everything that does not add value Waste can be found in any activity of the flow 13 Research Methodology Paper Outline The research methodology that is used throughout this work is the Design Science Research Method ology DSRM 5 DSRM aims to solve identified organisational problems by designing and evaluat ing IT artefacts It consists of six activities as seen in the figure 1 Figure 1 Detailed DSRM for this work Adapted from 14 This paper begins with a brief Introduction sec tion 1 about RPA and Lean It advances to the discovered Research Problem section 2 Section 3 is the related work for this paper In section 4 the Research Proposal and the Objectives are de tailed and its application is described in section 5 Demonstration Follows with an Evaluation of the results section 6 and concludes in section 7 2 Research Problem Nowadays RPA is being used as an almost instan taneous automation tool because it delivers quick results and doesnt involve changing the already ex isting underlying computer systems inside organisa tions 7 RPA developers dont usually care about optimisation 8 since that as long as it mimics the work that has been successfully done by human workers the project is sold But what if RPA tech nology was more present inside an organisations processes Two of the questions raised in this paper are how can RPA take advantage of a continuous process im provement approach and what is its impact To summarise what has been stated the main problem found is that RPA is being taken as a quick fix existing a lack of detailed processes analysis undervaluing its value to organisations when com bined with a process improvement approach such as lean 3 Related Work This section presents background information about RPA and Lean tools 31 RPA Life Cycle The currently RPA development model widely used is expressed in Fig 2 obtained from Anagnoste 2018 1 and Blue Prisms Guide 16 The main tenance phase is not represented Figure 2 Currently most used RPA development model from 1 First theres a gathering of the processes suitable for automation and then it proceeds with choosing one A specific RPA report from EY 4 enhances three key characteristics for a strong RPA candi date The actions need to be consistent the pro cess should be template driven and rulesbased A technical report from Symphony Ventures 2 utters an additional characteristic must be rawbased ie does not require human judgement element Next the goal of Process Reengineering is also to provide a highlevel analysis of the process solution the automation efficiency and estimation of effort 16 Following the project approval comes a detailed study where the RPA developer gets in detail with chosen process After having a functional documen tation and the project plan defined it is time for building the robot and test it There is no average time for the total time of an RPA project since it depends on the selected processes and its business complexity 32 RPA Tools Available in the Market For the past couple of years only Forrester Research 9 is doing research on RPA tools on the market and publicly communicate the results According to Forrester Researchs RPA Report from 2017 8 and from 2018 9 the market lead ers are Automation Anywhere Blue Prism and UiPath although currently there exists more than 15 tools 33 Lean 8 Wastes Lean tools are known for actively trying to reduce the following eight wastes 19 10 Transporting Excess Motion Waiting Overproduction Unneces sary Inventory Over Processing Defects and Skills 34 Lean Kaizen Kaizen is a Japanese word that translates to contin uous improvement Kaizen consists of small steps for each improvement whilst the traditional man ner mostly improves in substantial form causing high costs and affecting most people in the organi sation Regarding potential and benefits kaizen in vokes workers to participate in the upgrade from the 2 beginning unlike traditional improvement where the idea to change stays with the organisation top management until it goes live 11 35 Lean SMART Goals The appliance of lean tools to a system must have a reason goals must support that reason The term SMART Goals has been on trend for a long time and literature is abundant about its definition and usage Conzemius et al 2009 3 define SMART goals to be Specific Measurable Attainable Relevant and Timespecific 36 Lean Value Stream Management Value Stream Management is a process for planning and linking lean initiatives through data capture and analysis and consists of a guideline that a company must follow to achieve lean processes Tapping et al 2002 18 asserts that the guideline is suitable for office processes that is processes that dont pro duce a physical product but instead provide value to the final customer through a value stream com posed by work units 4 Research Proposal In this section the proposal objectives and the pro posal description are defined 41 Proposal Objectives This section is connected to the Definition of the Objectives for a Solution activity of the Design Science Research Methodology The primary objective this work aims to achieve is to understand how much more effective and effi cient Lean RPA1 is comparing to a standard RPA project To help solve that statement the following sub objectives must be achieved 1 RPA Tools Assessment In order to proceed to the next subobjectives its necessary to choose one tool to work with The two possible tools to work on are two of the current market lead ers BP and UiPath 2 Framework for Lean RPA This work defends that analysing and changing processes should be taken into account before the development of any RPA project This being said a frame work that joins both lean and RPA lifecycles must be created 3 Collection of RPALean RPA Results Gather results of applying both RPA and Lean RPA either in simulation or realfield in order to compare both approaches in terms of process ing time 1Lean RPA is now the created term for RPA when com bined with a lean approach Accomplishing these three subobjectives will provide enough results to answer the primary ob jective 42 Proposal Description This section is connected to the Design Devel opment activity of the Design Science Research Methodology The proposal model is represented in Figure 3 This model helps to visualise how the solution un rolls following the previously written three sub objectives The trigger Desire to Change is represented only to enhance a beginning It exhibits the sense of change and innovation that is necessary for pro cesses improvement projects If this trigger is ne glected then the following information is meaning less since it is based on continuous improvement Figure 3 Proposal model 421 RPA Tools Assessment Criteria The criteria for assessment was based on the criteria that Le Clair et al 2018 8 proposed but also on ISOIEC 25010 characteristics 6 that were mapped to RPA Functional Suitability Functional Appropri ateness Thirdparty Integrations Functional Efficiency Time Behaviour Error proofing Mechanisms Usability Previously required programming skills Overall Usability and available Training courses Reliability Robustness Security Robots deployment Scheduling and Logs quality 3 Maintainability Reusability of modules Portability Performance of the program on different information systems IS and perfor mance of robots 422 Lean RPA Framework Improvement Cycle The Improvement Cycle is detailed on the Figure 4 Each one of the four steps has a set of activities below that must be completed before moving on to the next step Those activities should be done in the order theyre presented 1 Process Selection The first step has activities are very centred on lean principles whilst the decision point starts tak ing RPA into account Lean suggests that defining a value stream VS is the initial point for change Next processes that belong to the VS should be sorted and one must be embraced to further mapping When mapping processes its fundamental to be simple yet detailed Everyone involved in the pro cess should be able to understand and identify waste tasks with at least one the eight wastes of Lean If the chosen process requires mainly cognitive actions even after identifying waste to be removed then its not suited for RPA and the first step should be repeated Otherwise its viable to continue to step 2 2 Process Study This step starts with the RPA Analysis activity Its necessary to assert what RPA can or cannot do in the process based on the current functionalities it offers Once that is finished its time for defining kaizen plans Kaizen is the correct choice here since it incrementally changes the process to consolidate with the RPA solution The next activity is mapping the future state Mapping the future state will help developing the robot implementing change to the process flow and having the new process documented Furthermore after having some documented processes improved by the improvement cycle it might become effort less to find similarities between processes and start standardising it Actions that should be redirected to the robot must be stated in the lowlevel map Before the implementation its also indispensable to define SMART goals for the project Without goals one could not understand if it was successful or not The goals should also be documented to be revisited later in step 4 3 RPA Project Development Its the step where the process flow changes should occur and the robot should be developed This step is on constant repetition so the process doesnt stop working another reason for kaizen plans After some iterations of the improvement cycle the developers might notice some similarities be tween processessubprocesses If thats the case this situation deserves an analysis to see if the used methodology can become a standard for the organ isation While implementing the robot standard opera tion procedures SOP and Simulations should be used to observe and test the development These activities avoid advancing to step 4 with errors and performance issues 4 Maintenance Review After the RPA Project Development its nec essary to review the goals For this one should compare the obtained results with the goals de fined in step 2 If the project achieves those goals the project goes live and a maintenance activity to that project shall be supported from now on If the project doesnt comply with the goals from step 2 its necessary to go back to step 2 and check what went wrong Finally the cycle ought to continue since lean proclaims for continuous improvements 423 Collection of RPALean RPA Results Once the improvement cycle has had enough pro cesses the results for each selected process should be gathered through simulation for RPA approach and realfield logs for Lean RPA approach 5 Demonstration This section connects to the Demonstration ac tivity of the DSRM 51 RPA Assessment Results The methodology used to gather information about the selected RPA tools was based on the analysis of the characteristics presented in technical reports by the vendors Forrester Wave reports 9 8 partici pation on official online learning courses communi cation with other professional RPA developers and more than one year of experience developing robots for professional and recreational purposes The results are stated below and Table 1 briefly summarises what has been assessed and the prices for each tool per year It is assessing UiPath without Orchestrator no scheduling and additional costs and without the additional cost of acquiring the development environment for more 3k Table 1 shows that its safe to assert that UiPath is more adequate for entrylevel Its cheaper and possesses an active online community and free train ing However being BP more robust its more adequate for processes dealing with audits frauds situations etc It takes longer for a beginner to 4 Figure 4 Lean RPA Framework Improvement Cycle Table 1 Analysis summary for the two assessed tools Parameters BP UP Functional Suitability 3rd Party Integration Time Behaviour ErrorProofing Mech Programming Skills Overall Usability Training Courses Robustness Scheduling Logs Quality Reusability Performance on Df IS Price per robot 18k 6k learn due to its complexity Its price is the triple of UiPaths price but that can be reasoned with the Scheduling feature which allows robots to work 247 for processes that require it 52 Lean RPA framework Improvement Cycle The demonstration for this subobjective was per formed inside a private Portuguese bank The infor mation gathering for the following subsections was done not only using indirect and direct observation but also reading documented processes and ques tioning employees about their routine at work RPA was already being used to tackle backoffice tasks In this demonstration processes will be modelled using BPMN due to its convenience on understand ability and on simulations 521 1 Process Step 1 Process Selection Choose Value Stream Firstly it was necessary to assess the value in the bank banking services For every bank service its necessary to open an account The ABC team is the responsible and seemed a good place to start The bank has one clear objective reduce as many FTE from ABC team as possible The ABC team deals with new accounts and any alterations to accounts that may occur Choose Process There were some chores stated as obnoxious and no one ever wanted to do it The most cited process was the DMIF Forms2 Fulfil ment This process consists of receiving the notifi cation that some customer updated hisher DMIF form and proceed to manually update every form in every account the customer is inserted Map Current State Identify Waste The pro cess is mapped in Figure 5 The activity in purple is the identified waste Searching for new DMIF alter ations can be considered as over production since sometimes there are no alterations for that day and time was just wasted in that activity Nonetheless the whole process is mechanical and requires zero cognitive actions This is directly mapped to the 8th waste of lean skill Once settled the project advanced to step 2 process study Step 2 Process Study RPA Analysis Define Kaizen Plans Search ing for new DMIF alterations is something RPA can easily do However the robot was not able to screenscrape the new answers from the form This 2DMIF Forms are forms that ask customers about their financial experiencie Its obligatory to be fulfilled for every account and always uptodate 5 Figure 5 Original DMIF Forms Fulfilment process was not a satisfactory solution since it couldnt re move any waste Knowing RPA current capabilities and limitations the IT department was contacted to know if there was a database with the answers It was affirmative but would take some time to give access to the robot The process could not stop working whilst being changed Thus the defined plan was to start with the implementation described previously This way workers would not waste time searching because the robot would send an email with the information if there were any alterations Whenever IT would give access to the database the rest of the project could continue Map Future State Detailed Study of the Pro cess The future state is mapped in Figure 6 The activity Search for new DMIF alterations is still considered waste since there might be no new alterations since the last check However it is now attributed to the robot which takes seconds to do Also if there are new DMIF alterations it instantly proceeds to do it Define SMART Goals The robot must be able to perform one instance of the process from the searching to one DMIF alteration in less than five minutes Its also intended that the DMIF updates are performed in every customers account within twentyfour hours of the original alteration from the customer Additionally after the final project goes live the ABC team can no longer interact with the normal flux of the process Once settled the project advanced to step 3 implementation Step 3 RPA Project Development Implement Kaizen Plans Develop Robot Cre ate SOP and Simulation Observe Performance First the ABC team started getting emails from the robot about any DMIF alterations With those emails ABC used the information to update the DMIF As soon as the IT department gave access to the database the rest of the project was developed within one week Step 4 Maintenance Review Figure 6 Finished DMIF Forms Fulfilment process with Lean RPA Review Goals The first and last goal were met as detailed later in next subsection For the second goal it was also met by merely running the robot every 24 hours Thus there was no need to regress to step 2 Go Live Maintenance The process went live This generates logs that are useful for further main tenance and to obtain results for evaluation The first process using the cycle was then seen as completed and more processes were on the queue for RPA 522 2 Process Step 1 Process Selection Choose Value Stream Choose Process Map Current State The DMIF Forms Fulfilment process reminded ABC team of another search they need to make every day unfailingly The bank possesses three types of passwords for accounts When a customer requests for missing password or wants to change one a request is gen erated on the banks intranet The ABC team is then responsible for searching new requests and re turning them to the customer Identify Waste The identified waste on the pro cess was the skill waste since searching for pass word requests is direct and mechanical This process requires cognitive actions to at tribute passwords however searching for passwords requests is direct and rulesbased Therefore this subprocess progresses to step 2 Step 2 Process Study RPA Analysis Define Kaizen Plans Since the robot was able to search effectively for the DMIF Forms Alterations in the last process this sub process is the same procedure but for another type of request Map Future State Detailed Study of the Pro cess Sending Passwords will now begin its flow with the robot The robot sends an email for ABC with the requests from the past twentyfour hours 6 and the team continues the original flow accord ingly Define SMART Goals This subprocess should take no longer than one minute Additionally every request should be completed in the following twenty four hours on weekdays Step 3 RPA Development Project Implement Kaizen Plans Develop Robot Cre ate SOP and Simulation Observe Performance This subprocess was developed straightforwardly since it only needed to run a query obtain the results build a table with the necessary informa tion and send it to ABC by email This can be come a standard methodology for upcoming similar processessubprocesses Step 4 Maintenance Review Review Goals The first goal was met as stated later in the next subsection The second goal was also met By sending an email to ABC with all necessary details its easier for the team to know how to carry on Go Live Maintenance The process went live This generates logs that are useful for further main tenance and to obtain results for evaluation The second process using the cycle this time sim pler but useful was completed 523 3 Process Step 1 Process Selection Choose Value Stream Choose Process Accord ing to the first process and bank intentions ABC team is still the chosen value stream The ABC team consists of 4 members and every working hour there must be at least 2 members validating docu ments This is because every operation related to accounts requires it Thus the next process should be one to relieve this burden One of the most prominent processes in the bank is the Accounts Opening This subprocess con sists of a Portuguese customer entering in one of the bank agencies and proceeding to the open an account with the commercial person After this the commercial person sends all documents back to ABC team to validate basic information on them and to open the account Map Current State Identify Waste The origi nal state is mapped abstractly highlevel in Figure 7 In the Figure 7 the waste is represented in pur ple in the Commercial lane Contact customer for missing documents and Receive missing docu ments are activities of over processing and defects waste The commercial person is obligated to re peat the action of digitalising documents and send ing them to ABC The orange activities are the skill type of waste Figure 7 Original Accounts Opening in the Agency process Despite the marked waste activities theres also inventory waste The bank keeps every document even whats not necessary to open an account or the customer stops sending the remaining documents being idle Although the commercial part requires cogni tive actions validating documents is straightfor ward since its only necessary to read the document and check if it conforms with the rules Therefore the project advanced to step 2 Step 2 Process Study RPA Analysis Define Kaizen Plans Applying straightforward RPA development to the process would entail ABCs verification for every account due to todays OCR low reliability This would not be practical nor reduce any waste However check ing documents and sending the process to conclu sion is accessible for RPA and reduces some working time on ABC team members Nonetheless the process flow needs modifica tions Map Future State Detailed Study of the Pro cess The future state is mapped in Figure 8 The most significant change is in the Commercial lane instead of the Robot The commercial checks with the customer for the necessary documents before starting signing up If the customer doesnt bring the mandatory docu ments then its informed by the commercial person on the missing documents and how to proceed with the signing up Afterwards the robot just needs to open the account by checking the documents and send the opening process to conclusion As seen the ABC team is now spared from this process Define SMART Goals This process needs to re duce its average time by half to spare one FTE Additionally after the final project goes live the ABC team can no longer interact with the normal flux of the process Once settled the project was meant to advance to step 3 However it didnt progress any further The Lean RPA results presented in the next section are obtained through simulation 7 Figure 8 Accounts Opening in the Agency process with Lean RPA 53 Collection of RPALean RPA Results To collect RPA results the only RPA processes were designed and simulated for twentytwo days one month This was also used to collect Lean RPA results from the 3 Process The Lean RPA results were collected on the field for one month Every time the robot would run a certain process or subprocess it generated a log with what was done and the processing times 531 DMIF Forms Fulfilment Results Table 2 shows the obtained results from the DMIF Forms Fulfilment process The written values are the number of minutes wasted on the process per weekday Keep in mind that the average of DMIF alterations is four but this is not a reliable value since the process is initially triggered by the cus tomer and therefore is not constant Table 2 DMIF Forms Fulfilment process time re sults in minutes per day without waiting time Team Original RPA Lean RPA ABC Avg 27 195 0 Min 15 0 0 Robot Avg 011 076 Min 008 008 532 Searching for Password Requests subprocess Results Table 3 shows the obtained results from the au tomated subprocess Searching for Password Re quests This table doesnt include a column Pro cess w RPA because investing in automation for a subprocess as simple as this one would not be re warding This was only automated because of the lean approach Table 3 Searching for Password Requests sub process time results in minutes per day Team Original Lean RPA ABC Avg 15 0 Robot Avg 1 533 Account Opening Results Table 4 shows the obtained results in the real field original and simulation RPA Lean RPA from the Account Opening process The written values are the minutes wasted on the process per every new account Minimum values are ignored for Lean RPA due to the absence of work repetition Table 4 Account Opening process time results in minutes per new account without waiting time Team Original RPA Lean RPA Com Avg 32 32 36 Min 275 275 ABC Avg 125 787 0 Min 725 5 Robot Avg 176 07 Min 176 6 Evaluation This chapter is connected to the Evaluation ac tivity of the Design Science Research Methodology It follows some of the evaluation criteria proposed by 15 such as Efficacy Generality and Validity from the Goal dimension and Utility Fit with Or ganisation Harnessing of Recent Technologies from the Consistency with organisation criterion from the Environment dimension 61 1 SubObjective RPA Tools Assess ment The bank had already bought a RPA license UiPath However the banks reasons correspond to the results of the assessment It was due to its price and overall appropriateness and usability There fore the assessment reaches the validity criteria as it matches with the bank choices 62 2 SubObjective Lean RPA Frame work Regarding generality it determines how general is the artefact Its necessary for the proposed frame work to have a broader range of possible processes to act on to The framework was validated in the bank and there were no barriers in any phase How ever its not possible to guarantee that the evalua tion is entirely positive on this criterion because it needs to be tested on more realities 8 Table 5 Average time difference in hours in one month for each process without waiting time Process Original Process Process w RPA Process w Lean RPA 1 DMIF Forms Fulfillment 99 715 028 2 Searching for Password Requests 55 037 3 Accounts Opening in Agency 3263 3053 2691 After applying the solution in the bank and get ting positive results from the teams involved the bank will continue using it for its processes im provement Consequently the framework passes the following evaluation criteria utility and fit with organisation Finally it arose from the necessity to approach a new technology RPA in a different way than the companies have approached Thus the framework harnesses recent technologies 63 Primary Objective To recall the primary objective for the proposal was to understand how much more effective and efficient Lean RPA is comparing to RPA Regarding the efficacy criterion which is the de gree to which the artefact produces its desired ef fect the table 5 supports the primary objective evaluation Table 5 represents the three processes presented in Demonstration section 5 on monthly aver age processing time difference The following ta ble doesnt take into account the waiting time each process has Its assumed one month equals to 22 working days that the new accounts average is 20 per day and the average of DMIF alterations is 4 per day The sum of all the time saved using Lean RPA instead of normal RPA equals to little over six working days of one FTE for one month 2739 Although only three processes were studied the bank naturally has way more manual and rulesbased processes more specifically the backoffice department Continuing the au tomation journey reducing FTE will be organic to the bank Its adequate to assert that in these cases Lean RPA was more efficient and more ef fective than RPA Efficient because the processes reduced their processing time and effective because changing processes taking into account RPA current capabilities will reduce any runtime mistakes 7 Conclusions Using Design Science Research Methodology this work proceeded to explain the problem found with the typical RPA implementations and objectives Being RPA a lightweight technology it is a quick au tomation that may fix some problems but its value is being undervaluing by organisations To try to prove it a primary objective has been defined to understand how much more effective and efficient Lean RPA projects were in compari son to RPA projects Primarily it was necessary to assess an RPA tool Blue Prism and UiPath were used and put to test It was concluded that although BP is more robust and reliable UiPath is the most correct choice for RPA starters due to its functional appropriateness and price Secondarily to obtain Lean RPA results creating a framework to follow its steps was essen tial The framework has incremental activities aim ing for the participation of everyone involved since the beginning of the project To culminate after applying the first two sub objectives the outcome was evaluated with the gathered results The framework was applied in one bank for three processes and the results were mostly positive Despite the organic time reduction that RPA projects already secure every process had a significant reduction on time comparing RPA and Lean RPA It was also successful concerning its util ity since the referred bank has continued following the activities Concerning the two asked questions in section 2 Research Problem How can RPA take ad vantage of a continuous process improvement ap proach and What is its impact Lean RPA takes advantage of a continuous improvement approach by automating and changing processes and sub processes that would not be approached by simply RPA projects since it doesnt fit the typical pro cess choice The studied processes were of easy and secure development and implementation whereas other complex processes would take way longer Due to its price fast development and effectiveness Lean RPA projects can change the way processes flow inside an entire an organisation with the right improvements and right thinking on the processes modelling and execution It does not leverage the overall cost by much since it uses basic ideas and takes advantage of internal teams help that already know the processes and workflows To conclude the DSRM the activity Communi cation is covered by the respective dissertation and this paper that communicates the results 9 71 Limitations This work was limited in terms that it should be tested more and on different realities Additionally the proposed Lean RPA framework is not a stand alone tool 72 Future Work The Lean RPA framework needs to be evaluated for its understandability for the users Future efforts are also required for an improved framework such as merging Lean with Six Sigma Acknowledgements Id like to thank my parents for the support Thank you to my supervisor To my friends thank you for never failing to make me laugh To Pedro thank you for always believing in me And finally thank you to everyone that I met through the development of this work References 1 S Anagnoste Setting Up a Robotic Process Automation Center of Excellence Manage ment Dynamics in the Knowledge Economy 62307322 2018 2 I Barkin Best Practices Guide Roadmap to Robotic Process Automation Technical re port Symphony Ventures 2017 3 A Conzemius J ONeill C Commodore and C Pulsfus The Power of SMART Goals Us ing Goals to Improve Student Learning Solu tion Tree Press 2009 4 B Diepeveen J Matcher and B Lewkow icz Robotic process automation Automa tions next frontier Technical report Ersnt Young 2016 5 Hevner March Park and Ram Design Sci ence in Information Systems Research MIS Quarterly 28175 2004 6 H W Jung S G Kim and C S Chung Mea suring software product quality A survey of ISOIEC 9126 IEEE Software 2158892 2004 7 M Lacity and L Willcocks What knowledge workers stand to gain from automation Har vard Business Review pages 17 2015 8 C Le Clair A Cullen and M King The For rester WaveTM Robotic Process Automation Q1 2017 Technical report Forrester 2017 9 C Le Clair G ODonnell W McKeonWhite and D Lynch The Forrester WaveTM Robotic Process Automation Q2 2018 Technical re port Forrester 2018 10 J Liker The Toyota Way McGrawHill Edu cation 2003 11 A Manos The Benefits of Kaizen and Kaizen Events Quality Progress 4024748 2007 12 A Manuel Robotic Process Automation Uma analise comparativa das solucoes actuais 2017 13 T Melton The benefits of lean manufacturing What lean thinking has to offer the process in dustries Chemical Engineering Research and Design 836 A662673 2005 14 K Peffers T Tuunanen M A Rothenberger and S Chatterjee A Design Science Re search Methodology for Information Systems Research Journal of Management Information Systems 2434577 2007 15 N Prat I ComynWattiau and J Akoka Artifact Evaluation in Information Systems Design Science Research A Holistic View PACIS 2014 Proceedings Paper 23116 2014 16 B Prism Lifecycle Orientation 2016 17 M Rußmann M Lorenz P Gerbert M Wald ner J Justus P Engel and M Harnisch In dustry 40 The Future of Productivity and Growth in Manufacturing Technical Report April BCG 2015 18 D Tapping and T Shuker VSM for the Lean Office Productivity Press 2002 19 J P Womack and D T Jones Lean Thinking Banish Waste and Create Wealth in your Or ganization volume 5 Simon Schuster 1995 10