iManagementBrazil Ltda.

View Original

Case Study: Process Mining & GenAI - the road will be longer than expected - or not?

Preface

For several years, we have periodically carried out various projects with iMB.Solutions Ltda. for a European system supplier to the automotive industry in Brazil and Argentina. As part of a global strategy to structure supply chains according to newly defined criteria, the strategic focus on the American continent has shifted to near shoring, combined with friendly shoring. The USMCA economic area in Mexico was selected as the operational test laboratory for this.

In order to realign the supply and value chains, the processes were to be made fundamentally transparent. This was done using process mining, which we had already applied at various other client locations in South America. What was new this time was that a GenAI was to be used specifically to evaluate the event logs in process mining. Our main task was to operationalize the new approach and to document and make transparent the experiences gained in daily use. In the near future, this pilot project may become a firmly established project management tool.

Project Briefing

Industry: Automotive supplier; system supplier; location Mexico

Our project function: Project Manager

Topic: Reorganization and transformation project Near Shoring USMCA

Turnover p.a.: not released

Number of employees globally/Mexico: not released

The project mission was executed during Q1 2023. Due to the grace period for the publication of the business case, the case study can only be published now.


Related Blog

The reader is recommended to follow the link below to find out more about process mining:


What was the assumption behind combining process mining with GenAI?

Process mining, supported by GenAI, can extend the application possibilities of process mining. For example, GenAI can improve process discovery by automatically generating process models from event logs. In addition, GenAI can also help with process improvement by automatically generating optimization suggestions. The combination of process mining and generative AI can therefore provide a powerful and flexible basis for comprehensive applications in process mining.

However, the classic applications of process mining quickly reach their limits in reality, especially when it comes to autonomous or highly automated optimization approaches. We have already had to make this experience in various projects over the last few years. It was therefore clear to us that traditional process mining was quickly reaching its time and cost limits. For this reason, our client had recently increasingly resorted to technologies from the field of GenAI and machine learning (ML) in order to expand the application possibilities and success potential of process mining.

Traditional approaches in process mining include process discovery, process conformance and process enhancement. In process discovery, the event data is evaluated and the actual process model on which the actual business processes are based is derived. In process matching, the actual model is compared with a typical target model in order to identify deviations in a first step and then to localize and explain them. During process improvement, the existing actual model is expanded, adapted and improved in order to achieve an efficient process flow.

process mining concept practiced by iMB.Solutions

The Process Mining GenAI Bot in Operations

Process improvement with GenAI can be carried out in various ways. One option is for GenAI to automatically generate optimization suggestions based on the existing process data. These suggestions can, for example, aim to eliminate bottlenecks in the process, shorten throughput times or improve the quality of the results. With this approach in particular, we were able to see very quickly that considerable productivity gains could be achieved.

Another approach is for GenAI to generate new process models based on the existing data. These models can then serve as a starting point for process improvement, for example by revealing weak points in the process or suggesting alternative process flows.

However, it is important to note that the use of process mining software solutions based on GenAI and ML usually requires in-depth IT and process mining knowledge. Although we have very profound knowledge of process mining, this approach requires a high level of support from trained IT specialists. The project approach must be approached with great sensitivity, as the time and budget limits of IT specialists are very quickly exhausted.

Project Management

To set up such a project, the following steps must be carried out:

  • Select the appropriate process mining software that supports or integrates GenAI applications, e.x. such as Pega Process Mining, Celonis or IBM Process Mining.

  • Definition of the goals and requirements of the project, such as improving process efficiency, quality or flexibility, reducing costs or risks, or increasing customer satisfaction, i.e. rigid project controlling.

  • Collection and preparation of event data from the various source systems, such as ERP, CRM, SCM or MES, which map the relevant processes. Cleaning and curating the data can usually only be done manually and is very time-consuming.

  • Analysis of the actual processes with the help of process mining in order to visualize and understand the process models, workflows, performance and problems. This should be done in the first step using the traditional method in order to bring the entire project team up to the same level.

  • Generate new process models or variants using GenAI applications based on the analyzed event logs that meet or optimize the defined goals and requirements.

  • Evaluation and validation of the generated process models or variants in order to check and compare their feasibility, effectiveness and robustness. It is essential that this is done in an extended team in order to allow all aspects of the process to have their say.

  • Implement and monitor the selected process models or variants in the target systems in order to realize and measure the expected improvements. Project controlling must be updated here and check the step very closely.

Advantages of a Process Mining GenAI Bot

  • The ability to continuously optimize business processes and adapt them to changing real conditions almost in real time by benefiting from the creativity of GenAI applications. The bot becomes a real assistant.

  • Reduce manual effort and human error in process modeling and improvement by leveraging the automated and data-driven capabilities of process mining.

  • Increase transparency and understanding of business processes by using the visual and interactive representations of process mining.

Disadvantages of a Process Mining GenAI Bot

  • Dependence on the quality and availability of the event data that forms the basis for process mining and GenAI applications. If the data is incomplete, inaccurate or outdated, the results of the project may be compromised. This applies to both the traditional and GenAI approach.

  • Complexity and challenge to understand, control and trust the GenAI applications, which may not always produce the desired or expected process models or variants. It can be difficult to explain or justify the logic or intent behind the processes generated. I.e., such projects need to be set up as contextual, holistic projects. The pure technical explanation approach is not sufficient; the entire project must be considered and explained in context.

  • Need to manage the organizational and cultural changes that go hand in hand with the introduction of new or changed processes. There may be resistance or skepticism from the stakeholders involved who are used to the existing processes or have other interests. This mistrust can be increased by the use of GenAI among employees. The contextual intelligence of the project or interim manager is therefore of fundamental importance.

The Business Case for Our Client

The business case of such a project depends on the specific circumstances and objectives of the organization in question, but in general we can draw the following conclusions.

  1. The costs of the project include purchasing or licensing the process mining software, integrating the source and target systems, training and consulting the project staff, performing the process analysis, generation, evaluation and implementation, and ongoing maintenance and monitoring of the processes.

  2. The benefits of the project include the savings that can be achieved by improving process efficiency, quality or flexibility, reducing costs or risks, or increasing customer satisfaction. These can be measured in the form of key figures such as throughput time, error rate, customer satisfaction, turnover or profit.

  3. The profitability of the project can be calculated using indicators such as return on investment (ROI), net present value (NPV) or internal rate of return (IRR), which express the relationship between the costs and benefits of the project. This must be defined in the client's project controlling before the project is started.

GenAI Software in Process Mining

Some of the current GenAI solutions for process mining are presented below. The list does not claim to provide a complete overview, as the market is currently developing dynamically.

Pega Process Mining

This software makes it possible to use generative AI models such as OpenAI's ChatGPT to propose new process models or variants based on the analyzed event data and improve process performance. The generative AI models can be used as a front-end interface to facilitate process analysis and improvement.

Pega Process Mining is a software that enables companies to analyze and optimize their business processes. With the help of AI-based process mining, users can gain insights into their processes and identify optimization opportunities. Pega Process Mining is a user-friendly solution that enables users of all skill levels, from business users to data scientists, to understand where their key process optimization and automation opportunities lie.

With Process Mining and the Pega Platform, organizations get an easy-to-use solution to uncover process inefficiencies and implement optimization initiatives within a single, seamless solution.

Celonis

This software offers a feature called “Process Automation Discovery” that uses GenAI algorithms to generate new process models or variants based on the analyzed event data to support process automation.

The GenAI algorithms can also simulate and evaluate the impact of the proposed process changes. Celonis is a German company that offers process optimization software. The software uses process mining and AI to analyze and optimize business processes. Celonis offers a user-friendly solution.

IBM Process Mining

This software uses GenAI technologies such as Generative Adversarial Networks (GANs) to generate new process models or variants based on the analyzed event data to increase process efficiency, quality or flexibility. GenAI technologies can also ensure the robustness and adaptability of the generated processes.

Comparison IBM Process Mining vs. Pega Process Mining

IBM Process Mining and Pega Process Mining are both software solutions that help companies analyze and optimize their business processes. Both solutions use GenAI-based process mining to provide users with insights into their processes and identify optimization opportunities. Both solutions are user-friendly and provide an easy way to uncover process inefficiencies and implement optimization initiatives within a single, seamless solution.

However, there are some differences between the two solutions. According to our usage, as users, we found IBM Process Mining easier to use and better suited to their business needs than Pega Process Mining.

On the other hand, as users, we were very impressed with the setup of Pega Platform and the management of processes. When it comes to the quality of ongoing product support, we preferred IBM Process Mining. In terms of features and roadmaps, we prefer Pega Platform.

To summarize, IBM Process Mining and Pega Process Mining offer similar functionality, but there are differences in usability and features. It is essential to consider the specific requirements of your organization to find the best solution for your needs. I would not give a clear preference.

Learnings

Below are the scenarios and results that emerged from the combination of Process Mining and GenAI; the internal management implications and conclusions have not been authorized for publication by our client.

  1. Optimization and Increased Efficiency

    By implementing Process Mining and GenAI, business processes can be significantly optimized. GenAI definitely helped to generate new process models and to quickly identify and eliminate bottlenecks and inefficient processes. These were primarily localized in the reduction of throughput times, an increase in the quality of results/meaningfulness and a general improvement in process performance. These efficiency gains in turn led to cost savings and improved customer satisfaction.

  2. Challenges and Delays

    There is a possibility that the project may take longer than expected. This is mainly due to the fact that none of us really have in-depth experience in the integral use of GenAI. Some of the reasons identified for this were manifold:

  3. Data Quality and Availability

    The quality and completeness of event logs is crucial to the success of process mining. Incomplete or incorrect data could slow down the process and affect the results. This applies to both traditional and GenAI-supported project implementation.

  4. Technological Complexity

    The integration of GenAI into process mining requires in-depth IT knowledge and possibly also extensive training for virtually all employees involved. The need for highly specialized IT specialists could lead to delays, especially if these resources are scarce. The amount of training required was initially underestimated.

  5. Cultural and Organizational Resistance

    Changes in processes may encounter resistance within the organization. Managing these changes requires time and careful planning to increase stakeholder acceptance. Such resistance can occur in principle, but is currently being intensified by the use of GenAI, as there is still a high level of mistrust.

  6. Successful Implementation and Expansion

    As the implementation was successful and the expected improvements were realized, the pilot project could serve as a model for other locations and projects. Our client could adopt process mining and GenAI as a firmly established project management tool and apply it in other areas. This would increase the company's competitiveness and strengthen its position in the market.

  7. Unforeseen Results and Adjustments

    The combination of Process Mining and GenAI led to unforeseen results. This was due on the one hand to the very in-depth process analysis and on the other hand to the results generated by the combination. For example, we often found that the GenAI-generated models and optimization proposals did not always have the desired or expected effect. In such cases, the project team had to be flexible enough to make adjustments and develop alternative solutions. We learned very impressively here that the use of GenAI in process mining also served us well as an intensive process design thinking assistant.

  8. Long-term Innovation Benefits

    In the long term, our client could gain an innovative edge through the integration of GenAI in process mining. The ability to analyze and optimize processes in near real time could lead to continuous improvement and adaptation to changing market conditions. This could also improve the company's ability to innovate and react quickly.

Conclusion

The outcome of this business case depends heavily on the quality of the data, the technological expertise of the team and the ability to manage organizational change. While the project took longer than originally expected, it also offers the opportunity for significant improvements and long-term benefits. The keys to success lie in careful planning, the right choice of tools and technologies, and effective change management. It is safe to say that the use of GenAI significantly increases the depth of understanding, dramatically expands the ability to communicate results across the organization and significantly accelerates the flexibility to think and test new processes.

Observation

The project was carried out 1.5 years ago. Due to the grace period for the publication of the business case, the case study can only be published now. We know from other project missions that the environment for GenAI applications has changed massively in the meantime, including in process mining, and that the applications have developed at an extremely rapid pace. Therefore, the statements made in this case study might also be re-evaluated in new projects.

See this form in the original post