Marlon Dumas

Prescriptive process monitoring techniques recommend actions in an ongoing case of a business process to maximize its success rate. Different techniques have been proposed that focus on the efficiency and precision of recommendations. In contrast, little attention has been given to presenting the outputs of techniques to end users. In this study, we design an interface for prescriptive process monitoring outputs following the design science research methodology. As an artifact, we develop a web-based tool Kairos. We evaluate the tool with end users (operational managers and process analysts). Based on the findings, we derive suggestions for designing prescriptive process monitoring interfaces.

Since the early 90s, the evolution of the Business Process Management (BPM) discipline has been punctuated by successive waves of automation technologies. Some of these technologies enable the automation of individual tasks, while others focus on orchestrating the execution of end-to-end processes. The rise of Generative and Agentic Artificial Intelligence (AI) is opening the way for another such wave. However, this wave is poised to be different because it shifts the focus from automation to autonomy and from design-driven management of business processes to data-driven management, leveraging process mining techniques. This position paper, based on a keynote talk at the 2025 Workshop on AI for BPM, outlines how process mining has laid the foundations on top of which agents can sense process states, reason about improvement opportunities, and act to maintain and optimize performance. The paper proposes an architectural vision for Agentic Business Process Management Systems (A-BPMS): a new class of platforms that integrate autonomy, reasoning, and learning into process management and execution. The paper contends that such systems must support a continuum of processes, spanning from human-driven to fully autonomous, thus redefining the boundaries of process automation and governance.

Predictive Process Monitoring is a branch of process mining that aims to predict the outcome of an ongoing process. Recently, it leveraged machine-and-deep learning architectures. In this paper, we extend our prior LLM-based Predictive Process Monitoring framework, which was initially focused on total time prediction via prompting. The extension consists of comprehensively evaluating its generality, semantic leverage, and reasoning mechanisms, also across multiple Key Performance Indicators. Empirical evaluations conducted on three distinct event logs and across the Key Performance Indicators of Total Time and Activity Occurrence prediction indicate that, in data-scarce settings with only 100 traces, the LLM surpasses the benchmark methods. Furthermore, the experiments also show that the LLM exploits both its embodied prior knowledge and the internal correlations among training traces. Finally, we examine the reasoning strategies employed by the model, demonstrating that the LLM does not merely replicate existing predictive methods but performs higher-order reasoning to generate the predictions.

Business process simulation is an approach to evaluate business process changes prior to implementation. Existing methods in this field primarily support tactical decision-making, where simulations start from an empty state and aim to estimate the long-term effects of process changes. A complementary use-case is operational decision-making, where the goal is to forecast short-term performance based on ongoing cases and to analyze the impact of temporary disruptions, such as demand spikes and shortfalls in available resources. An approach to tackle this use-case is to run a long-term simulation up to a point where the workload is similar to the current one (warm-up), and measure performance thereon. However, this approach does not consider the current state of ongoing cases and resources in the process. This paper studies an alternative approach that initializes the simulation from a representation of the current state derived from an event log of ongoing cases. The paper addresses two challenges in operationalizing this approach: (1) Given a simulation model, what information is needed so that a simulation run can start from the current state of cases and resources? (2) How can the current state of a process be derived from an event log? The resulting short-term simulation approach is embodied in a simulation engine that takes as input a simulation model and a log of ongoing cases, and simulates cases for a given time horizon. An experimental evaluation shows that this approach yields more accurate short-term performance forecasts than long-term simulations with warm-up period, particularly in the presence of concept drift or bursty performance patterns.

In business processes, activity batching refers to packing multiple activity instances for joint execution. Batching allows managers to trade off cost and processing effort against waiting time. Larger and less frequent batches may lower costs by reducing processing effort and amortizing fixed costs, but they create longer waiting times. In contrast, smaller and more frequent batches reduce waiting times but increase fixed costs and processing effort. A batching policy defines how activity instances are grouped into batches and when each batch is activated. This paper addresses the problem of discovering batching policies that strike optimal trade-offs between waiting time, processing effort, and cost. The paper proposes a Pareto optimization approach that starts from a given set (possibly empty) of activity batching policies and generates alternative policies for each batched activity via intervention heuristics. Each heuristic identifies an opportunity to improve an activity's batching policy with respect to a metric (waiting time, processing time, cost, or resource utilization) and an associated adjustment to the activity's batching policy (the intervention). The impact of each intervention is evaluated via simulation. The intervention heuristics are embedded in an optimization meta-heuristic that triggers interventions to iteratively update the Pareto front of the interventions identified so far. The paper considers three meta-heuristics: hill-climbing, simulated annealing, and reinforcement learning. An experimental evaluation compares the proposed approach based on intervention heuristics against the same (non-heuristic guided) meta-heuristics baseline regarding convergence, diversity, and cycle time gain of Pareto-optimal policies.

This paper addresses the following problem: Given a process model and an event log containing trace prefixes of ongoing cases of a process, map each case to its corresponding state (i.e., marking) in the model. This state computation operation is a building block of other process mining operations, such as log animation and short-term simulation. An approach to this state computation problem is to perform a token-based replay of each trace prefix against the model. However, when a trace prefix does not strictly follow the behavior of the model, token replay may produce a state that is not reachable from the initial state of the process. An alternative approach is to first compute an alignment between the trace prefix of each ongoing case and the model, and then replay the aligned trace prefix. However, (prefix-)alignment is computationally expensive. This paper proposes a method that, given a trace prefix of an ongoing case, computes its state in constant time on the length of the trace using an index that represents states as $n$-grams. An empirical evaluation shows that the proposed approach has an accuracy comparable to that of the prefix-alignment approach, while achieving a throughput of hundreds of thousands of traces per second.