[MUSIC] Today's guest on the course is Marcus Henriksson. Hello, Marcus and thank you for joining us. You took the initiative to introduce an AI system in your organization. How did you motivate your initiative vis a vis the leadership of the company. >> So we built our case towards our executive management team in a number of different ways. You have to remember here that this was in the at the end of 2016 and beginning of 2017. So this was very much early days and I'd say that nearly 100% of all the decisions were business case or are y driven? So at this time we conducted a pre study to analyze the automation potential within the using different cognitive capabilities such as RBA OCR and machine learning and this showed a huge potential of cost savings. And the reason why [COUGH] this huge potential was that there were very large volumes of repetitive a rule based task which required no or little judgment. And even test that only took about 5 to 10 minutes per case became highly interesting since these cases could be multiplied by 40,000 companies. So, a process that saved maybe 10 minutes per case could add up to 6500 yearly recurring hours. So this strong business case gave us the executive management support and resources necessary for further in Quantum Asian, in fact, it was rather matter of managing expectations, which in many cases were unrealistically high. My department support at least 20 to 25 countries to help them define their automation strategy and to build their ability. And many of these countries initiated their automation without a business case and nearly all struggled to get the funding and scale their automation from their proof of concept automation. So in other words, a very classical example of the importance to involve top management. >> All right, so then where did you suggest that is the starting point of the optimization process for other people that want to imitate a little bit or be inspired by you? Which resources do you think are needed first? >> A centralization of no too low judgment task had already been initiated in the company to a shared service center and these tasks processes meet most criteria for what constitutes a suitable automation. For example, simple rule based repetitive tasks in large volumes and this is where we started now, the start is important. Many makes mistakes here and so did we. It is very tempting to take on the process on the list with the absolute highest savings understandably, however, not so wise if these processes are the most complex, which is very often the case. In true lean and adult spirit, it is without a doubt, much better to start small with simple process that can be put into production in a short time. This gave us an opportunity to learn and improve it enabled quick benefit realization and time to market which boost energy and commitment to our levels of the organization. And this approach with quick wins also helped convince those who initially were skeptical. So continues communication and broadcasting, showing how much savings we realized were absolute key. There are so many lessons learned here in and many seems obvious in hindsight. Remember to break down processes and long processes into smaller pieces but make sure to minimize the number of handovers between humans and the digital employees. It is wise to take on similar or nearby process as this allows for re use have already automated tasks. Back to the question, what researchers did you you need first? Approximately 1 2 3 full time equivalents can be sufficient as a starting point. This can then be scaled up as the number of automation candidates increase and the most obvious competence is the subject matter expert in the business process area to be automated. This can be an auditor, tax lawyer administrator. The next competences needed were the developers and business analysts who are skilled at the relevant cognitive technology and the business process for engineering. In addition to to them, it's also important to have access to architectural competencies who has a holistic business and technical understanding, for example, who can say when different capabilities are best suited. Should you use RPA OCR or versus integration. I recommended lean agile way of working and organization with a heavy emphasis on continuous learning where you can encourage everyone to dare to commit mistakes and to learn shorter durations using retros etcetera, helped us faster in this mindset. >> My next question is if you wanted to change an existing process or instead you wanted to create a new one from scratch, should these two alternatives be planned differently or are they the same? What did you say? >> These are very relevant questions and something we've learned at a very early stage. Regardless of whether you automate existing processes or new process, it's important to design the process based upon the principle automation first. So in other words design the flow that the automation can be so that that the automation can be as robust and efficient as possible. I should point out however, that with existing processes you are not always able to do a complete business redesign as these processes are in a context with other process, in an existing organization with existing roles and responsibilities and the automation can impact this whole. Sometimes it's worth the effort. Sometimes it's not. With a completely new process, however, greater freedom is given which enables you to take on a more disruptive approach. So how can we, in the smartest way possible simplify this flow. These workshops are highly creative and a lot of fun. >> Okay, I see. Then I would like, I would like to ask you also something about the factors that you would say we're really decisive for the success of the automatization process. Was there something really making it or breaking it? >> There is much to be said about this and one is asked to summarize key factors for successful automation in many ways it boils down to the same components as with any change initiative. And this is true also here. Top management, support funding and common objectives are absolutely crucial. Change management and communication impacted parties need to be involved whether they be management IT, HR employees consultant, they should be involved early on in the initiative. And in addition to the more generic factors, I also want to point out a few more specific to automation. It's extremely important to have a continuous candidate generation pipeline where you have an ongoing identification and the vetting of potential automation candidates. I also want to stress that lean agile where you're working comprising of business technology teams and by this meaning we're not working as auditors and customers. We have the one team spirit. And you also need an automation strategy is showing the direction the ambition and selection of appropriate capabilities, and the Mantra Automation 1st. So automation robots, machine learning AI it's all about the data. I cannot stress the importance of the data quality and the master data. >> All right, so now moving on to the human factor. Did automatization create any resistance within the organization and what did you do to reduce this resistance if it happened? >> So did the automation create any resistance in our organization and what did we do to reduce this? Well the answer is yes. Very much so. In spite of all that we already know about how to implement changes in order to avoid resistance. We hugely underestimated this in the beginning and made a number of false assumptions. One, we underestimated people's fear of losing their jobs, since which a large extent automated were conducted by hourly employees and consultants who didn't have a long term plan with a job. We didn't think they would have any objections. We were wrong. The fear resulted in obstructing behaviors where they even worked against the automation. So what did we do to handle the situation? We decided to involve these people in the development to a much larger extent the expression human in the loop is relevant not only to ensure humans and robots and digital employees interact efficiently, but also to make sure humans are involved and understand the purpose and the business development and the changes necessary. We train people in what automation is, what a digital employees regardless of if this was an RPA or an OCR robot or another form of machine learning and AI solution. We simply explained all the benefits with the automation. Yet at the same time we were transparent and honest about the fact that work will be replaced and by doing so, we were able to have an active dialogue with the people who feared and/or opposed to change. Another factor which complicated complicated our collaboration in the beginning was that we realized that we had conflict in goals in the organization. The shared service sent was measured on the number of hours centralist annually. And initially they were not allowed to include the hours they were automated. So they were actually literally punished when the automation was successful and vice versa. So just by including the automation and emphasizing automation first hugely improved our collaboration. And then working in lean agile business technology teams not only strengthened our joint cooperation and efforts, but it also enabled us to handle problems, conflicts, disagreements upfront at an early stage in our daily stand ups and retros. >> All right, thank you, then, my final question is about large versus small and medium sized companies should one thing differently when introducing AI systems depending on the size of the organization. Should small companies automate eyes their processes at all? >> Absolutely, without a doubt, all companies must automate their business in one way or another, respective the size of an organization everyone must ensure that their business is running the most efficient, automated and digitalized way possible. It's easy to come across as overdramatic and exaggerate the consequences for those who don't keep up. However, automation digitalization is to a large extent in hygiene factor today to get to be able to stay in the game at all. Since automation and digitalization in general is an investment into the future large companies should and must create this capability in house. This competence can be used to either streamline a business and cut costs or to generate new business models, products or services. My recommendation to smaller companies who in many ways have the benefits of not being burdened by legacy systems and structures is to initially seek external help from someone who has long hands on experience of automation and digitalization within a holistic view of the world. If you don't mind me say so, there are many out there who sells one size fits all kind of solutions don't buy it. But with the right helps more organizations can gradually recruit the relevant staff necessary as they move forward and progress and with time they will learn what needs to be done and what can be left untouched for all sized organization as mentioned earlier. Start, small, lean, agile manner and remember to have a long term vision and a lot of patience and grit. I've seen too many pilots and proof of concepts end up in the proof of concept or pilot graveyard due to unrealistic expectations and no direction. Don't become one of them. >> All right, excellent. Thank you so much for joining us today. Once again, Marcus, thank you bye bye.
