New approaches to project management: AI as the key to 38% more efficiency

June 13, 2024 – Reading time: 6 minutes

The use of artificial intelligence is becoming increasingly important in today’s business world. Artificial intelligence (AI) in project management (PM) aims to automate workflows and improve efficiency and quality. However, according to recent surveys, 67% of companies still use manual data entry. Companies that use automated interfaces account for less than half. In addition, 49% are aiming for comprehensive automation. These figures illustrate a broad willingness and great potential for the use of AI in project management. (Source: The Project Group [TPG])

A model analysis we carried out shows that AI support has particularly great potential in resource management, project planning and updating project data. Through the effective use of AI, time-consuming tasks such as reporting and plan updates can be automated. This not only leads to an increase in efficiency, but also to an improvement in quality, for example in the creation of business cases, project communication or risk identification. The study shows that the effective use of AI can already save 38% of project management costs. This means that a project manager working 40 hours per week can free up to 15 hours per week. The free capacity can be used for activities that increase the quality of the project and lead to better project implementation. A more detailed breakdown of the savings potential can be found in the following figure.

Source: INVENSITY Estimates

In addition, the quality of decision-making can be increased in many project tasks through a data-based approach. Ultimately, relieving project managers of communicative and repetitive tasks allows them to focus more on project goals and controlling, which ensures the long-term success of the project.

There are three basic approaches to introducing artificial intelligence in companies. In the following, we look specifically at the area of project management – however, the approaches can be applied analogously to all areas of the company, such as research and development or administration. Which approach is the right one for your company depends on your specific goals and requirements. We would be happy to work with you to define and analyze your goals in order to determine the best approach for your company.

In-house development: The development of in-house AI solutions enables the targeted adaptation of AI to specific problems and requirements. In addition, absolute control over all processed data is guaranteed. However, this requires considerable development effort, as the development, training and optimization of a company’s own algorithm is necessary. Due to the associated costs, this solution is only economical in special cases, for example for applications with strictly confidential data.

Use of PM tools with integrated AI: When using PM tools with integrated AI, the tools are integrated into the company’s infrastructure. An example of this is the implementation of a new intelligent tool for resource allocation. These tools are ready to use and offer automatic data collection. The disadvantage, however, is the limited adaptability, as customization is only possible within the framework of the respective tool. In addition, the introduction of a new tool is time-consuming and requires employees to be convinced of the solution and receive standardized training.

Embedding proven AI language models in the company infrastructure: In our view, this is currently the most promising solution. By using proven language models, integration into existing tools can be achieved and the introduction of a new tool in the organization can be avoided. This increases the flexibility and adaptability of the solution.

The first important step in the targeted use of an AI language model in project management is to provide the required data for the corresponding use case. In many cases, this is done using machine-readable information from PM tools or documents. For example, effort data in hours from a ticket system or cost data from a spreadsheet. Input from the project manager, such as information on the project for creating a schedule, is also possible. For optimum results, this input must then be made available to the language model via a prompt – i.e. a structured instruction. The prompts are optimized in advance for the specific application. This is because a prompt for creating project schedules is structured differently to a prompt for determining measures. The embedded language model then outputs an initial result. As a rule, however, this is not yet machine-readable. Language models do not output usable information, such as an introductory text like this: “Here you will find a tabular list of the project tasks […]”. The machine-readable information can be filtered via an interface and checked for the correct format (CSV, continuous text, value) via verification. This is initially a formal check. The subsequent content check during validation ensures that the result makes sense in terms of quality. For example, via critical parameters that are defined in advance. One such parameter could be the deviation between best/normal/worst-case estimates for an effort estimate for a task. If the deviation from the value range is too high, the result is returned to the language model and corrected. The output values can also be compared with historical real data. In the case of a project schedule, for example, with the actual duration of similar projects in the past.

In summary, this approach creates a fully controlled and use-case-specific infrastructure that makes it possible to generate verified and validated results. The user benefits enormously as a seamless connection is created between generative AI, data interfaces and existing PM tools (such as Jira, MS Project, Polarion, …).

Our goal is to understand your specific requirements and develop customized solutions that can be seamlessly integrated into your existing infrastructure. We help you identify and implement the best solutions, while ensuring that ethical standards and data protection requirements are met. In addition, we ensure that our solutions are ISO/IEC 42001 compliant. We also understand the importance of transparent communication and collaboration with the works council, especially when it comes to the use of AI and the associated impact on employees. From the initial consultation to the training of your employees, we will guide you through the entire process to ensure the successful introduction and use of the new technologies. If you have any questions, please contact Marc Bollmann.

Contact person

Resources