Best Practices Manual

This manual was developed within the AI-ImpactSK project to translate research findings into a practical process that enterprises can use when deciding how to introduce and further develop artificial intelligence. It is intended for enterprise management, process owners and teams responsible for information technology, data, security, compliance and human resources.

The manual draws on the empirical and analytical outputs of the AI-ImpactSK project. Its evidence base includes three separate quantitative surveys conducted within two waves of data collection, in-depth interviews with enterprises, expert webinars, segmentation, statistical and econometric analyses, comparative assessments, and relevant regulatory, ethical and methodological frameworks. Their synthesis made it possible to identify recurring conditions, barriers, risks and best practices associated with AI implementation in business practice.

The key findings are translated into a ten-step decision process. It guides an enterprise from defining a specific problem through assessing readiness, selecting a suitable use case and designing a pilot, to measuring results, managing risks, setting rules for use and deciding on the next course of action. Each step ends with a decision gate that helps determine whether the enterprise can proceed, whether the solution needs to be modified, or whether implementation should be paused or terminated. The process is therefore sequential, but not strictly linear.

How to Use This Manual

The manual is practical and methodological in nature. It does not replace legal, security, sector-specific or certification assessment of a particular AI system. The scope of controls and documentation should be adapted to the specific use case, the type of data, the degree of autonomy of the solution and the associated risks.

Best Practices Manual