As we progressively explore the multitude of functionalities and applications of artificial intelligence, exciting projects keep coming up. We’ve seen how AI can increase efficiency in customer care, and how predictive tools based on data mining are paving the way for a new generation of public services. Kicking off this year, a research team led by a professor from the University of Tartu (EE) will now find out what AI can do in order to improve business processes for companies.
The team, under the guidance of Professor in Information Systems Marlon Dumas, has recently been awarded a European Research Council (ERC) Advanced Grant for a spell of five years. The project falls into the framework of an investment of €540 MLN made by the European Union into cutting-edge research. Professor Dumas and an international pool of collaborators aim to develop a tool to automate the detection of the root causes of ineffective performances in a business process. Moreover, the Process Improvement Explorer (PIX) they’re developing will take corporate efficiency one step beyond, by designing potential solutions to intervene right on the dysfunctional link in the chain.
Admittedly, the subject is not one of the easiest to handle – and that’s what we are here for. Professor Dumas took the time to have a chat with us and respond to our questions on features, purposes, and practical cases where PIX can prove all its added value. At the meeting point between research and business development, innovation comes at the service of companies and entrepreneurs to increase efficiency and provide new opportunities. Almost needless to say, “once again”.
What is the Process Improvement Explorer (PIX) and what are its main features?
PIX will be an Artificial Intelligence machine that will scour through data extracted from large-scale information systems. Its aim is to identify issues that affect the performance of the business processes of an organization. For example, think about a handful of customers in a given region who are experiencing delays in receiving the products they have ordered. Naturally, these customers will start visiting the company’s website often to check the status of their delivery.
PIX will detect that these customers are experiencing a delay and will trace down the reason for these delays. It might be, for example, that a supplier is waiting for missing information to fill out a customs clearance document. PIX will detect it as the root cause, and it will propose ways of preventing this problem from happening in future.
What is the current situation in the management of business processes? What are the main advantages that PIX could bring in terms of improvement opportunities?
Issues such as the one mentioned above often go undetected, especially when they only affect a handful of customers and only in some circumstances. When an issue affects a large percentage of customers, or when it occurs over and over again, the managers will become aware of it. They will instruct their analysts to look into it. The analysts will investigate the issue using existing analysis methods and tools, for example using so-called ‘process mining’ tools. With some effort, they will eventually find the root cause and they will make changes in the business process to address it. For example, they might propose to add an item in the checklist used by employees in the company’s warehouse, asking them to enter the required customs clearance information for every product that contains liquids.
However, a lot of issues do not occur often enough to get to the level where managers will worry about them. They are dismissed as exceptions because they occur rarely. The problem is that these rare exceptions are often so numerous that, after a while, they become the norm. In the era of highly digitized processes, companies spend less than 20% of the effort dealing with 80% of normal cases in their processes. The remaining 80% of the effort is spent on dealing with exceptions. The problem in modern organizations is not how to handle the normal cases, but rather how to handle tons of exceptions. PIX will allow companies to go after thiese long-tail of exceptions.
PIX is not only a monitoring and diagnostics tool, but it also has a prescriptive and predictive function. How do AI and data mining enable these purposes?
Yes, correct. PIX will not only tell you that some of your customers are experiencing a problem after-the-fact. It will also find the 10-12 customers, out of thousands, who are going to experience a given problem in 1-2 weeks’ time. There is already technology for predicting problems. But the challenge that PIX will address is how to make these predictions explainable and actionable. Yes, now I could tell you that your customers will, at some point, experience a problem. But what if we were able to know what you can practically do to effectively prevent this problem from happening? This is the challenge that PIX will address. We call it prescriptive process monitoring – it’s not only about predicting, but also prescribing efficient and effective solutions.
How does the AI think within the framework of PIX and business management? Will it follow some specific lines of action?
Back in the 70s and the 80s, several management theories emerged that describe all the different types of problems that may happen in a business process. For example, there’s a management discipline called the Toyota system, that tells you that there are seven sources of inefficiencies in a process. These methods are commonly used by analysts to manually improve business processes.
PIX will design artificial intelligence algorithms that will mechanize these methods. These algorithms will look at every transaction recorded in a company’s information system, on the lookout for inefficiencies. Incorrect practices will also be detected, and whether these may lead to customer complaints down the road.
The project is boosted by European Research Council (ERC) funds. Why these funds are important, and what impact can frontier research make on our lives?
Nowadays, researchers spend a lot of time running behind funding. The research we do is largely determined by the funding we get. My research has a lot of practical applications, so I regularly get funding to do short-term projects with companies. But companies think in timeframes of 1-2 years – and sometimes even shorter. If my team keeps doing these projects, we will make lots of small improvements in existing tools. But after a few years of doing that, we would only be making incremental improvements. We would just keep applying what we already know.
The ERC is different. It funds fundamental long-term research. With this funding, we can think in a 5-10 years’ timeframe. We can explore crazy ideas. When you look at what existing process mining tools do, PIX looks like a crazy idea: How can you automate something that, using existing tools, takes analysts days to find? We think we can, and the ERC funding allows us to take the risk to pursue our ideas.
If we fail, we’ll go back to doing short-term projects in 5 years time. If we succeed, we’ll be able to commercialize a system that can find things that current tools simply cannot, no matter how hard you shake them.