Sarah Fane 0:00
Hello ladies and gentlemen and a very warm welcome to today’s webinar brought to you by Shared Services link and sponsored by high radius. Today we’ll be looking at GE Digital Transformation guide for shared services. My name is
Sarah Fane and I am the Head of Research at Shared Services link. And I’m delighted to be joined today by Vivek Thakral , who’s the director of artificial intelligence at GE Power, and Elaine Nowak, who’s the Director of Product Management at marketing at high radius. So let’s have a little bit of context for today. And what we’ll be looking at in this webinar. Digital Transformation is certainly one of the hot phrases of 2019 in shared services. And it really is the next wave of disruption that we see is fundamentally changing the way shared services operate. It is transforming shared services from being what we call cost killers, to really becoming value drivers. However, we’re seeing actually, there’s quite a large gap. And most shared services don’t quite have this TD transformation roadmap that will take them on this journey. We recently saw the Hackett Group Survey which said about 97% of finance professionals so that digital transformation will fundamentally change their operations. But only about a third of people said they had this Dziedzic roadmap in place to meet those goals. So really, what we’re hoping today is that you can start to build that roadmap for yourself. And we will be exploring GS automation journey to understand the impact of the digitization on the workforce and explore practical use cases of artificial intelligence for order to cash specifically, and shared services. So with that, I would now like to hand over to Vivec, who will be sharing some more information on GE Powers journey.
Vivek Thakral 1:40
Thank you, Sarah. Good morning. Good afternoon. Good evening, everyone. Thanks for joining this webinar. My name is Victor Thakral. I work with GE Power has been 14 years with this company and today, I lead the delivery of artificial intelligence solutions to solve some of the business problems to cash management sourcing, finance, as well as operations. And I’ll take you through a glimpse of the overall GE Power business. We are a global manufacturing company focused on manufacturing large equipment or turbines to produce electricity globally. Today, GE Power produces electricity for 1/3 of the world’s population. We are also heavily engaged with the government to do the nuclear solutions, as well as provide services to our customers and also providing digital solutions to track and monitor equipment. If you think about the global footprint of our company, we are across 100 different manufacturing sites, primarily North America, Europe, Asia, Latin America, as well as Africa. Today, we have more than 24,000 installed base that provide electricity to 1/3 population of the world. And majority of our suppliers are across different regions as well. We have more than 2500 IT systems including enterprise systems like ERPs, Salesforce PLM, and various data lake solutions. For a company like GE, why would we focus on digitization, it is definitely becoming a norm, where you want to be looking at digital transformation to solve business problems to a shared services organisation where you have a central data lake that gives you a single source of truth and insights to your business operations and business needs. As well as systems that can connect to various functions. Starting from your engineering to manufacturing to sourcing to logistics, you need a shared service organisation, which has an end to end visibility, how you receive an order from the customer and then how you mature it and also how you recover your cash. That’s why shared services are extremely important for a large company like GE and digital transformation is just a fuel to the shared services organisation. Now, if you think about the overall vision, why this is important. In today’s world, large companies definitely are investing heavily into digitising their business operations. But the big question is, are you doing digital? Are you really being digital? So we conducted a survey with a lot of our industry peers external to GE understanding how they are progressing in their digital journey. A lot of times a lot of leaders, they mentioned that 70% of the leaders said they lack the skill to adapt to any digital skills. Or a majority leader also said they are either building the skills to digitise a business or they’re learning how to do so. The way I look at GE as a company. We think we are somewhere in the middle. We are investing heavily into our key enterprise solutions and architecture to digitise our business needs and operations as well as same time. We are making sure the data on it Information is available to the key strategic decision making leaders. So we are in the middle of a journey from doing digital to really being digital. So yes, the way I would suggest AI to look at is it has two components. No AI is possible without human beings, and no AI is possible the doubt machines. So combination of human and machine makes AI. And this is how we are also implementing the technology within GE. It is not for the sake of implementing a fancy technology, but we are solving business problems to AI. As the first step, what we do is we we look at the business process, we partner with the business leaders, like a sales leader or an engineering leader manufacturing leader, logistics and finance to understand what are some of the business problems or operations they’re trying to solve? And are there any dependencies, or any errors, or many manual effort that they’re trying to reduce in order to become more efficient, and effective, based on all the processes or the leaders that we work with, we put together a blueprint or a roadmap to do automation. And these automations as I said, humans and machines. So human beings are the subject matter experts who understand these processes. And machine is there is no one solution that fits for every process or the problem, we architect the solution. Thinking about the middle layer that you see here, that artificial intelligence, we think about do we need optical character recognition to scan your PDF documentation or scrape some data or make it voice enabled, that you can search a specific piece of information like a sales order number, purchase order number, customer name, one supplier name in that document. So every solution is architected. Based on what is the business problem that you’re trying to solve today, and GE Power, we do majority of these AI solutions except the physical robots. As of today, machine learning is also playing a huge role. And analysing what has happened historically in that operation or process by analysing the data that we have in our lakes, as well as predicting what could happen in future predicting customers behaviour, if they would definitely go ahead with the order, or they will cancel it or they will modify it, or even predicting if, if the supplier will deliver the raw material on time. So machine learning is playing a huge role in predicting a lot of business insights. With that we have a very strong governance process in place, a team of experts and automation experts who are closely partnering with the business to implement these technologies, we have taken a more strategic approach in putting an enterprise architecture together. And I’m talking in terms of Auto Cash in specific if you think any standard enterprise architecture is three layers, system of records, that’s where your real business value lies, and most of your operations are done. These are your PLM systems or ERP systems where your maximum number of business users across different functions. These systems take multiple years to implement, and to run the business and they last long five to 10 years before you can make any business change, you need to think of our system of records providing that visibility and access to the information and the middle layer, the system of differentiation. These can be deployed in a shorter period of time, but they give you data analytics or insights to make business decisions, dashboards and any kind of insights that comes from upstream or downstream processes. On the top layer, these are more of you can implement them faster, like predictive analytics or Salesforce opportunities that you capture from a customer in those system. Those gives you quick and easy way of implementation and integrating with your system of records. Similar to this, what we have done in GE, we have put together an artificial intelligence architecture in line with the three layers that I just talked about the way we think about the system of record layer, robotic process automation, it fits well with the system of record, because these automations are literally using simple business rules to automate tasks. For example, automate creation of a purchase requisition, or automate creation of a purchase order or automate creation of an invoice or distribution invoice. So RPA is they are introduced as employees within the organisation. These employees access a system of record to perform a specific task and the middle layer. These are more of cognitive skills of AI where it is not just a bot or a robot just doing a task but also understanding and having a conversation with the end user. For example, if the user wants to know status of a purchase order. So they can simply type the question or speak the question, what is the status of my field. So when you parse this information, what is the status of my purchase order my relates to a persona, the person who logged in and was chatting, maybe a buyer. So the cognitive AI should be capable enough to understand who logged in and who’s asking for the purchase order and what all purchase order was created by this individual. From that they will retrieve all the information and give you back in either a chat box, or probably send you an email or give you a dashboard access. So those are easier to implement faster technologies. So you’re using different AI tech stacks to do so. And on the top layer, is machine learning. As I was explaining, these gives you meaningful insights into the business. If you want to change your decision or you expedite a decision, you want to cancel a purchase or, or you want to send a reminder to the customer to make payment on time. So machine learning gives you valuable insights. That’s how we have put together an architecture of AI. The way we have taken an approach of introducing automation within GE Power, they cannot be a roadmap, you cannot define a timeline or six months or one year or 18 months. To deliver certain kind of automation, we call it as a maturity model. The automation journey or automation has a maturity. So in the first phase, we give three phases for this maturity model. And the first phase we call it a initialised phase. We call it AI one that was primarily last year when we were understanding the best practices of AI in the industry. We partnered with Gartner, we partner with Big Four consulting companies, we partner with management consulting companies to understand how other industry peers are thinking about implementing AI. We spoke to 34 Different companies ranging from finance to insurance to industrial companies, even with the US government, customs, border security and IRS to at least share some best practices and understand the headwinds and tailwinds that could enable a company like GE to implement AI, so we did a lot of learning last year. Based on that we thought about, we need to start reaching out to our functional leaders to generate interest, motivate them to identify opportunities and create a backlog of different automation opportunities. From there, we looked at what would we need in terms of skill set within the company from a technical skill set to a process skill set to a development skill set to fulfil the need of AI. A lot of majority of the last year was went into setting up a world class organisation and infrastructure and frameworks of AI to build automations. We have a very strong governance council that evaluate every automation use case, what business problem is trying to solve is it trying to play on a standard process versus the process itself needs some kind of fine tuning? What will be the outcome of that process, whether it will give you better cash, whether it will give you better inventory optimization, or cycle time reduction. So it has to tie to a specific business goal. Our governance team does a lot of that. Last year, while we were establishing a learning the framework, we also focused on delivering fear of automation, which I’ll talk about in coming pages. But today where we stand, we have moved on to next phase, which is the eye to the impact phase, we are making an impact by putting together a top down strategy or a blueprint with our system of record leaders, ERP leaders, functional leaders, like CFO supply chain nearer plant managers, as well as operations leads to understand where are the pain points? What are the problems? And how do we address those? Based on the discussion, we conduct lean workouts, identifying the defects and then find the improvement opportunities, laying out the as his business process and suggesting a to b process with a combination of different AI technologies to RPAS to OCR to Chatbot. We configure blueprint and then we start building either bots or chat bots or machine learning solutions. We definitely have a very strong intake process so we go out there conduct roadshows educate our functional teams about what AI can do. It is not just going to disrupt your existing job roles, but it will definitely come and be your assistant a bot or AI solution will be your personal assistant doing repetative job for you. If you think about if a company hires and experience, chartered accountant or a CPA to do finance analytics and make financial decisions. You don’t want this individual to create general ledger entries, or create invoices or make phone calls to the customer. So what we do is we conduct roadshows to explain to our leaders that human beings were hired to do a specific job, let them do what they do, and do the repetative, monotonous, and rule based jobs to any automation solutions. So that’s why we are creating an ecosystem of how do you utilise human beings to be more efficient by leveraging the AI technologies that we are developing for them. So we today where we are as the ITU phase where we are very mature in terms of RPS and machine learning solutions as well. The next phase that I foresee is the ICT three phase, which is industrialise the more democratise you do this model, or you explain to the leaders a particular leader, the better it is, for any organisation ID, this is going to be enabling our existing workforce to take better decisions and be more efficient in what they do. The ID phase is more of an open source, kind of an AI framework or architecture, where we would enable a business functional owners to get access to AI technologies, modified tweak programme or even add rules in existing AI technologies to solve some of the business problems. So we are putting together a framework that will have capabilities like process mining, or cognitive services, even IoT, like a company like GE, we have a tremendous amount of fleet out there delivering the equipment, or tremendous instal base like gas turbines or windmills out there, that can be tracked to IoT and automation. So next phase is coming soon, where we going to scale our automation capabilities beyond the system of record is going to go out there.
Vivek Thakral 17:01
I would like to definitely share a little bit about the journey so far, how it all started, the little successes we had so far, and how we are maturing this to the next level. So if you think about why did we even do this, what was the need of doing automation, it all started early last year, when we had put together a vision with the CEO of the company talking about what are the top 1000 or 500 100 processes within the company that has manual dependency or that are labour intensive. So we did an analysis on all the processes across our manufacturing shops, or finance services, or field services. And we found there are more than 2500 processes that are done manually each day, and millions of hours are spent to execute these transactions. And about a half a million man hours are spent to rectify any errors that a human would be would have made by doing the original transaction. So from those 2500 processes, we narrowed down on the top 10. And those top 10 processes could be something like goods movement, between one warehouse and other, or creation of an invoice, customer invoice. It could be distribution of the invoice to the customer, or it could be creation of a journal ledger entry, or cash application cash collection dispute management. Those were some of the top 10 processes that we identified has got a lot of labour intensive and manual intensive jobs. So we spoke to a CEO we shared the analysis. And he said, let us just automate one process and see the impact. So if you see last year in March, we went live with our first automation called Max we name every automation we give them a unique employee ID. Every automation has a manager in the system that has an email id, each of their activities and logs can be tracked. So we introduced Max as a biller. Max his job was to create a customer invoice based on the approved sales order and approved purchase order making sure the data and that invoice the name of the customer the bill to information shipped to in the amount of payment terms, everything is accurate. Max started creating those invoices for North America billing team and started distributing to the customers on their either their portal or they were electronically transmitting to the customer systems. Max saved about one hour and 20 minutes per invoice to be created and distributed. But the volume was huge. There were more than more than 5000 invoices that were processed by only one team in North America. So we could have saved more than seven times And I was with one team only. We build that automation, we showcase the CEO, they acknowledged it, they saw the value in it, it went up to the chairman of the company, the board of directors of the company, they all acknowledged and that’s how it all the journey started. With on and on, we went live with all of and we had a big bang release last year at 11 bots in July today, where we spend, we have 54 automations live in our G power enterprise architecture. And these automation has saved us more than 100,000 man hours. We are not suggesting that these are specific human beings or FTS, we are suggesting we have given back 100 1000s of productivity time back to the company so they can better utilise the human workforce and make them more efficient and effective.
Vivek Thakral 20:48
So with that, I’d like to give you a little more detail in depth view what what a digital worker is and what a bot is. So, this is an anatomy of a digital worker, every digital worker as I said is introduced as an employee these are we consider them as contract employees with specific roles and responsibilities to do whether they have doing a desk job to create a requisition or invoice or they are communicating with the customer telling them to make the payments on time or they are communicating with the suppliers to procure the material within the agreed cycle times. So every automation we introduce has a name as an identity and they have motor skills to continuously do monotonous repetitive rule based tasks. So RPA is doing a lot of that for us. And then we are adding the speech capabilities. We are working closely with the vendors directly Amazon and Microsoft to use the natural language processing capabilities where you can give a voice based search command for example, if you have a mechanical engineer in a workshop installing a turbine, he or she could give a voice command saying what is the availability of part ABC in my warehouse in South Carolina, it will understand the question, run a query and give you a result and respond back saying this specific part you have availability is zero. Would you like me to order it? And in this response, you could say Yes, could you ask max to create a requisition for supplier ABC, then Mac’s will get triggered and do the procurement cycle. So those are the speeds level capabilities and more similar to that think about vision. We receive a lot of scanned PDF documents in the shape of purchase order or any instructions that an OCR technology can read it, scrape it digitise information, enter into ERP and just trigger the necessary downstream things. And finally, every automation not every I mean a lot of automation we do we think about adding analytical skills, where every transaction processed by an automation is tracked, analysed and predicted how this this transaction will optimise the entire downstream process. If bought has created an invoice, what is the confidence level that we will get paid on time by the customer record has got zero errors. So those kinds of analytical capabilities is what we’re implementing with automations.
Vivek Thakral 23:26
And now let’s focus a little more upon the overall order to cash automation. This is a blueprint that we typically formulate with the billing and the collections leader talking about what are they trying to achieve, this is the entire end to end lifecycle of an OTC how we receive a purchase order in a PDF from the customer. So we have an automation or a bot that receives that purchase order reads it scrape the information and reconcile it then we have another board that is creating a project out of the information from the purchase order which is part of the sales team. Similar to that we have created automations are part for the billing team sales team as well as cash management of cash collection teams, all the output of one bot becomes the input of another. So we are focusing on improving the data quality. So there is very less manual data entry or manual creation of transactions here because this is all automated relying on single source of truth relying on master data relying on the data lake that we have. So these automations are accurate business rule based and each of the output is being tracked and monitored by the billing and collection leader on top via a dashboard which has got machine learning capabilities. So the billing and collection leader can track and monitor every bot how they’re performing with the purchase order number being created and led into a sales order, the sales order matches with the invoice number and invoice number matches with the original purchase order amount, everything matches all the information as shown on a dashboard to be a lot more efficient in terms of entire photoessay process, definitely till date, we have a bunch of these automation in place running in production environment has reduced a lot of duplicate records, reduce a lot of repetative work or fixing any manual errors, and has helped us collect cash faster from a customer. So we have definitely reduced the past deals by millions. As of today, like we speak. I want to shift gears a little bit talking about how all overall automations will impact the workforce of future. Before I dive into any details, if you go to any search engine, and you start typing just two words, this type robots will or robots will or RPS will just start this type these two words, and certainly will start giving recommendation like robots will take over my job robots will destroy humanity, robots will disrupt the world. So you will get a lot of negative press and emotions out there from the internet. But if you really think about how automation can really help or impact the workforce is four different phases. And this is this is a theory also we are applying in GE we are monitoring each phase. So the phase one the way to think about is you may not be able to automate the entire process. In every process, there is human decision making needed you need these human beings. To do the negotiations with the suppliers, you need the human beings to maintain the relationship with the customers, you need human beings to do the mechanical jobs for that for which they were hired. So in the first phase, we have to identify what are the specific task that you need to automate through different AI technologies. That’s where majority of our automations are today. The 54 automation we have majority of them are task automation. So we are doing a lot of task automation. Once we we are mature in that space, you’re not thinking about how do you add more cognitive skills like conversation AI, you are having a conversation with a chat bot looking for your purchase order or invoice number. And if you do not find you should be able to give command to your conversation AI to create one for you by triggering a bot at the backend. So understanding what a human being is trying to say and triggering the necessary automated task to do the job is what his cognitive skills are. That’s why are we are investing efforts today as we speak. The next wave is definitely an ecosystem a partnership between a human being and a bot human beings will have bought other assistants, virtual assistants to do repetitive tasks and human beings are tracking and monitoring their behaviour to either fine tune the automation change your rules, understand the exceptions and taking the decisions to better use the automation. So it is a combination of both a human being and the part and the fourth phase is coming. It’s coming soon. Any organisation your think about not just human beings and organisation will have a combination of employees obviously you need them, contractors, automation of bots and different tech stack of AI how to use them. So that is a vision that we have also painted in GE so that is the future state or the end version that we also have a GE this overall organisation that I’m talking about. Think about from the from the bottom, you have process mining process mining techniques, gives you insights on the variances in the process. If biller A is creating an invoice in five hours without making an error, and the biller B is creating the same invoice in 10 hours, but he did five additional steps. So why variation. So with the process mining techniques, you could derive three outcomes from that. The first outcome is you are not using your standard billing process in the company. You probably want to activate the standard process of billing, you have to make it standard. Second outcome is you have the standard process. But these two individuals or the billers are doing it differently, which means you lack training. So you need to train these human beings and make them more equipped with bots or make them more equipped with better standing operating procedures to do the jobs right and do it efficiently within five hours. So you got to train them. The third recommendation could be you have this process enabled. You have your human beings trained properly Then why don’t you just automate. So the third recommendation is automate build bots. That’s what is shown on this screen as well. So think about a procure to pay organisation with multiple automations. One bot is creating your RFQ one bot is creating a purchase order one bot is talking about receiving. And all of their activities are tracked by a supervisory board which is a machine learning bot, analysing and predicting behaviour of your bot organisation. And finally, as machine learning bot is feeding into a control tower, which is a dashboard I was mentioning earlier, where you have all the insights for the human beings or the business leader to take the decisions that they should be taking. So this is this is our future state. I would say, as of today, my focus is on getting this army of bots together based on the outcomes we are receiving from process mining techniques. So we are putting an army of bots, then we will have machine learning to supervise them and then control our to give recommendations.
Sarah Fane 31:08
And with that, I would like to hand you over to Elaine Novak from HighRadius.
Elaine Nowak 31:12
Thank you so much, tremendously interesting information. From Vivek. I’m just gonna take a moment and talk a little bit about high radius, we had high rated, we actually take the digitization of the OTC process very seriously. Our latest release offers the artificial intelligence enhanced technology across our entire platform. But just to give you some background, we as a company we were founded in Houston, Texas, and we’ve been offering solutions for the accounts receivable and credit space for more than a dozen years, we are the only provider of an integrated receivables platform that provides end to end solutions for the entire order cast process. And we’ve infused that artificial intelligence into our product suite. So going back to where our expertise and our experience lies, that’s the foundation of what we did to create our cloud based products suites for the Receivables Management. And because we focused on the automation of this process, we have become the market leader, especially in the Fortune 1000 area in which we are ranked number one among accounts receivable solution providers. And just recently, in the last couple of years, we have secured a number of investments, that’s going to help fund our continued growth and expansion. And just to give you a sense of our size, we process upwards of 1 trillion in receivables annually for our customers 1 trillion that’s with a tee. And in terms of size as a corporation, we have about 900 employees globally. And we’ve got offices in America, in Europe and in Asia. And then looking at the kind of customers that we have within our organization. We have products that encompass revenues of all sizes for different companies, we’ve focused across various industries, you might find on the screen, you’ll see a lot of brand names that you are most likely familiar with them. And so in the course of working with these customers, we’ve actually had upwards of 650 transformation projects that we have conducted over the course of the years. And looking a little bit more in depth at our actual integrated receivables platform. We have five main modules. And again, we’ve in our latest release, we have the artificial intelligence that’s infused in our technology. In our credit cloud module, we’ll help you get your new customers on board and we help manage the ongoing credit risk for your existing customers with things like an automated online credit application, automated credit bureau integration, credit scoring and automated workflows. For our electronic invoice, presentment and payment, we provide our customers a platform for them to engage with their customers digitally. And we give them the option to view invoices view statements, make payments and log disputes. It’s a merchant branded portal with invoice templates, automated dispute management, and a payment gateway integration that’s available for our cash allocation cloud. We help automate the application of payments once the received and eliminate a lot of that manual task that Vivek referred to all of those tedious sort of administrative work that can get done by these bots, and by artificial intelligence and robotic process automation. So you’re able to data image capture, that’s done automatically, regardless of the format. It’s the ability of the invoices to be matched automatically. And to receive that straight through processing, using the artificial intelligence, and also the ability to recognize short pays and deductions and do the coding automatically have those short pays. And that leads into the deductions cloud, what you find is those customers because they’re not always paying in full, the cloud actually helps manage those short pays resolve the disputes faster. It has automated bots that go out and download claims information and backup information related to the deductions. It has automated trade matching. It also provides shortage and price variance analysis, and it also creates automated claim denial packages that can go out based on a set of criteria. And then lastly, we have the collections cloud, which helps organizations create strategies. It helps them to prioritize how their follow up will be with their customers. It automates notes, it automates reminders and the payment commitments within the system. It also provides automated dispute management and correspondence. So each of these five modules are mature solutions. They have deep functionality and robust capability built organically by high radius so they all have the same look and feel and the same ease of use. And when we designed our cloud platform, we built our architecture so that it would be highly scalable, that would have great resilience and would perform to very high standards. So that is one of the main reasons why we have more than 95% clients subscription renewal rates for our products.
Sarah Fane 36:19
Thank you for Vivek and Elaine for all the time and effort you put into this really great presentation. We hope you enjoy today and we look forward to welcome you next time.
