As directionality is acknowledged as an essential component of contemporary innovation policies, questions have been raised about how and by whom the directions of innovation policy are set. Despite considerable attention on this matter, contributions have not explored the policy process of directionality, nor examined empirical cases to study directionality in practice. Considering these gaps, this paper presents an in-depth case study of the Dutch Automated Driving initiative that was developed under a smart mobility agenda with transformative aims (2013–2018). This initiative was rapidly championed by policymakers, and the agenda geared almost exclusively to its development. To study the policy processes therein, we used an adapted version of the Multiple Streams Framework (MS) (Kingdon, 1984). MS suggests that directions of policy change are determined by institutional entrepreneurs who have access to policy venues. We found that these entrepreneurs used political strategies (e.g. framing, problem-solution coupling) to champion automated vehicles as a transformative technology. However, eventually the transformative potential promises were not kept, leading to policy failure. In contrast, entrepreneurs’ self-interests dominated the policy implementation phase. This paper suggests that more attention should be given in how directions set in the early policy phases can be kept throughout the policy process.