Information magazine of the Department of Industrial Engineering
Università di Trento
Information magazine of the Department of Industrial Engineering
Università di Trento
How is work organized within a company? Who makes decisions? Which activities need to be carried out, and in what order? And, most importantly, how can we represent and improve these processes in a way that is simple, clear, and easy to share?
These questions concern Business Processes (BPs): the set of activities and decisions that enable an organization to achieve its goals and create value.
As companies grow in size and complexity, and as they face increasing pressure to adapt to rapidly changing environments, the need for rigorous, systematic, and formally grounded methods for analyzing Business Processes becomes increasingly clear. This is particularly relevant given that Business Process Management (BPM) has so far not achieved the level of adoption that was originally expected. It is therefore important to have tools that make it possible to analyze, design, and modify business processes in a rigorous way, while also keeping them understandable and easy to share with business experts that don’t have specific competence with AI and Agents.
Traditional methods for analyzing BPs are based on flow diagrams. The process is mainly described by identifying individual activities or tasks and connecting them to form a directed graph representing the process workflow.
One of the most widely used standards is BPMN (Business Process Model and Notation), which represents processes graphically through symbols and connections.
Diagram-based methods are widely used both in academia and industry. However, they have several well-known limitations: they require specific technical skills to be created and interpreted, they can be difficult to decentralize across different business units, they can produce complex artifacts that are difficult to query, maintain, or evolve, and they can limit the direct involvement of non-technical business stakeholders.
For example, a manager or process owner may have an in-depth understanding of how a business process works without being familiar with process modeling techniques. This can create a gap between those who understand the business and those who manage the technology.
To overcome these limitations, the AGO (Agents, Goals, Objects) methodology was introduced. AGO adopts a declarative approach based on the conceptual modeling of business entities and processes using knowledge representation techniques.
Instead of starting with the sequence of activities, AGO starts from three fundamental elements:
More specifically, AGO is based on a formal framework that uses set theory and mathematical logic to analyze Business Processes, providing rigorous support throughout the analysis, design, and implementation phases.
During the analysis phase, AGO identifies and models these three fundamental business entities: objects, goals, and agents. Together, they provide the building blocks for reconstructing the entire workflow of a Business Process.
The idea is therefore not to focus immediately on the question, “Which activities are performed and in what order?”, but rather to start from the knowledge underlying the process: who are the people or systems involved, what are their goals, and which objects do they use or produce?
Based on this information, the system can automatically reconstruct the possible paths through the process.
The workflow paths are derived by reconstructing all possible routes from start to finish, without the need to explicitly model them as a sequence of activities.
The concepts identified during the analysis phase are stored in a knowledge base grounded in a formal theory. This makes it possible not only to automatically generate the complete workflow, but also to verify important properties of the process.
For example, the system can help determine whether a process is:
In other words, the goal is to move from a simple graphical representation of the process to a description of its underlying knowledge that can also be used by information systems. A BP is executed by interpreting the knowledge base, without the need to write code (no-code development).
In the next stage, the design phase, classes corresponding to the identified concepts are defined.
During the implementation phase, the schema of a database is then defined to store the instances generated during the execution of the Business Process, together with the operational logic that guides the agents’ actions.
This is where agentic artificial intelligence comes into play. Software agents can use the knowledge contained in the system to perform tasks, make decisions according to defined rules, and interact with other elements of the process.
One of the most interesting aspects of AGO concerns how process knowledge can be collected.
Within a company, information about a process is often distributed among different people and organizational units. Someone working in a specific area may have detailed knowledge of their part of the process, but not necessarily of the entire process.
One of AGO’s advantages is its decentralized analysis, which allows different business units to collect and contribute their local knowledge independently, supporting a clear separation of concerns.
The simplicity of the methodology also makes it possible for business professionals without advanced technical skills — such as managers who have direct knowledge of the processes involved — to participate in the conceptual modeling and design of Business Processes.
This can reduce their dependence on technical staff and help improve alignment between business needs and information technology.
Furthermore, storing process knowledge in a knowledge base makes it easier to update processes when business procedures change.
The development of the AGO methodology and its implementation through agent-based systems are the objectives of a joint project with the Institute for Systems Analysis and Computer Science “Antonio Ruberti” (IASI) of the Italian National Research Council (CNR) in Rome.
The project is currently focused on developing a proof of concept, providing an initial practical demonstration of the feasibility of the approach.
Once this phase has been completed, the goal is to develop a prototype agent-based system capable of supporting the application of the AGO methodology in real-world business environments.
The ultimate goal is to make business process management more understandable, collaborative, and adaptable, connecting the knowledge and expertise of people with the capabilities offered by agentic artificial intelligence.
