Responsible AI decisions at Flemish SMEs

VERA-AI is developing and validating a method to help a Flemish SME make an informed choice between traditional automation, an LLM workflow and an AI agent for a specific knowledge-unlocking task. This method explicitly assesses whether the AI and data should remain on the company’s own servers or run in the cloud, incorporates data protection from the design stage (privacy by design), and links each recommendation to an evaluation framework that measures whether the solution delivers value to the user in practice. We validate the method using three real-world SME case studies. The result is an open, freely available method with detailed sector-specific examples. The project aligns with the mission of the (Bio-)Tech11 Research Centre and with the ‘User-Centred Experiences’ priority area, which focuses on user-centred and reliable AI.

VERA-AI is aimed at small and medium-sized enterprises that wish to deploy AI but do not usually have their own AI team, regardless of their sector or digital maturity. Within the SME, the method is intended for those responsible for making technical decisions: the IT or digitalisation manager, or the external technical partner guiding the SME through the process. This decision is made before an AI solution is built. The method is therefore a technical decision-making framework, not business advice.