Artificial Intelligence supporting Process Development | QPR

QPR Blog

You are here


AI Supports Process Development - Understanding the Current State and its Problems

Artificial intelligence and machine learning change the world. In this blog, I open my thoughts on how the DEVELOPMENT of processes will work in five years, in 2023. All these ideas are already in some form so now is the time for process developers to wrap their sleeves!

Digitalization, artificial intelligence, machine learning ... business models are transforming, agile operators replace their sluggish competitors and information is the source of successful business! At these times when change is the only constant, the goal of process development is to make sure that the change leads to the right direction - do nothing or change nothing leads to loosing the battle for competition ! Here is my vision of how artificial intelligence will support process development.

Process Development in 2023
Basic steps in process development will remain the same. In this blog series, I structure process development to four stages as follows:
1. Understanding the current state and its problems
2. Identification of Root Causes
3. Making Intelligent Process Development Decisions
4. Executing the process development projects

Part 1: Understanding the current state and its problems
The traditional way to understand the current state of a business process is to interview people involved in the process. I have personally been involved in hundreds of workshops and interview situations facilitating discussions, asking the questions, modeling the business processes current status and building a shared understanding. However, it is no longer sufficient to only interview people. Artificial intelligence and advanced algorithms already have a major role in understanding the current state and its problems thanks to the following technologies:

Process Mining
Process Mining visualizes the current state of a business process as a flowchart. Analysis is automatically generated based on event data and logs stored in ERP systems. The method fits all business processes that use one or more information systems to support operations. Using Process Mining method, the process developer always gets an up-to-date visualization of the current state of the process, summary of process variations, bottlenecks, waste work and the lead time durations between any process steps. For process developer, it is a huge benefit that the busy business people are not required to join lengthly interviews in order to capture the process flow. As the process flowchart and analysis is automatically generated, the process developer can tell to the business people how operations are actually carried to complete the whole fact-based end-to-end process. My own estimate is that less than 2% of 1000+ employee organizations globally have been using process mining now in May 2018 and I personally expect that in 2023 already more than 50% of organizations having 1000+ employees will be using Process Mining method for at least some of their processes.

Text Mining & Speech Recognition
A large amount of business information is generated as spoken or written text during interaction between the customer and the supplier, as well as within the organization's internal encounters. In 2023, advanced speech recognition is able to convert spoken recordings into text format and further translate written text into a desired language, for example, everything could be stored as English text. Text Mining features will then be applied to this text-formatted material in such a way that extensive set of factors relevant for the process execution will be identified.

Categorization (Clustering, Unsupervised Learning)
Human beings are better than any other animal species in classifying things. However, the Artificial Intelligence and Machine Learning algorithms are already today better than humans in finding common attributes and classifying large data datasets. For example, in service business, the key ability for providing superior service to every customers is based on the service employees ability to categorize every customer into a relevant category and use that categorization as the basis for providing a service that is specifically suited for that customer. Al large Online Stores and search engines have a systems that give most relevant purchasing recommendations and ads based on the AI driven categorization or clustering engines. In 2023, the process developing expert will use AI-based clustering to categorize incoming cases (such as different customer groups, ordered products, business areas), process variations, resources involved in the process (like employees, machines, suppliers) and process results. Using clustering method, a spaghetti-looking chaotic process flowchart can be divided into several easy-to-understand clusters which each have a clear and meaningful process flowchart.

Identifying Trends
When change is the only constant, it is almost wrong to even talk about the as-is process descriptions as somekind of a static thing. Using process mining, we will see how the process was executed last year, last month, or last week. However in the changing world, the execution today is already a little bit different. Humans understand relatively well about large-scale changes, clear trends, and seasonal variations. Retail business is different during weekends than on weekdays, the night is different from the day, and summer season is different than winter season. Artificial intelligence, however, is not limited to such simplifications. AI understands trends for every day, time, new product launch, weather, TV program, politicians' election year, legistation change and the popularity trends found in Google Analytics. Artificial intelligence identifies trends in large data much better, more objectively, and more efficiently than human beings. In 2023, AI will provide process development with a summary of identified process trends at any desired time interval, say, once a week, a month, or a year. Artificial Intelligence finds trends in customer behavior, process activities, resources, end results, and lead time duration. AI will present the results visually in a format understandable by business people.

AI to understand the relevance of process observations
All the methods described above may give to the process developer either meaningful or insignificant information. In 2023, the artificial intelligence still does not fully replace humans from process development. Instead, the role of artificial intelligence is to produce revelant meaningful information to humans. Even as the latest technology, AI does not fully understand the nature and relevance of each finding in 2023, it will be able to estimate the relevancy of each process related finding based on the teaching data received from process developers. The algorithm works so that first the artificial intelligence produces some millions of observations from the process, prioritizes them on the basis of statistical significance and presents the top-10 observations to the process developer. Process developer then teach the AI machine learning system which observations are meaningful and which are insignificant. The feedback from the process developer teaches AI machine learning algorithms and soon the AI will get quite accurate in providing meaning relevant findigs for every separate human using the AI driven process identification system.


This was my part 1 about the topic "AI supports Process Development". Thank you for reading this far and as always I'm more than happy to receive comments,

Teemu Lehto

+358 40 546 0202

teemu.lehto [at]


Teemu Lehto
Vice President, Process Mining