Policymaking regarding Artificial Intelligence is threatened to lag behind, and the Dutch Ministry of Justice and Security stated its desire to catch up. The longer we wait, the wider the ‘gap’ will become between regulation and market development. In order to keep AI technology beneficial, a human-centered design approach should be key not only in AI development, but also in policy development. Instead of focussing on ‘what is possible’, we need to empathize with stakeholders to unravel the latent needs that define the public values of today. Along with the unfamiliarity of AI technology and its unpredictable consequences, this calls for a transition towards an agile government. This transition represents a shift from ‘assuming’ to ‘learning by doing’, from ‘is it safe?’ to ‘is it safe to fail?’, and from 'reactive' to 'proactive' policymaking. This thesis proposes proactive policymaking as a creative, iterative and future-oriented exploration. AI experiments should be framed as a priority vehicle for organizational learning. Experiments are used to empathize and unravel its desirability early in the process. Policy Provotypes and Minimum Viable Policies are created as first feedback loops prior to current feasibility studies. An ‘Desirability panel’, covering the experiences of a representative group of society, reduces blind spots and offers transparency. It facilitates short-line communication and alignment. Potential policy interventions transcend theoretical discussion as the demonstrated experiments provide a context for detailed dialogue. Policymaking will be less affected by the media's magnifying glass as stakeholder inclusion is at the heart of policy development. This vision calls for the need of a Creative, Empathic, Transparent, Inclusive, Future-oriented, Iterative, Experience-driven and Demonstrative approach. Therefore I present the acronym: CERTIFIED. The CERTIFIED approach covers a theoretical framework for agile policymaking. In essence, it is about the ability to adapt quickly by providing a window for feedback by the public early-on in the process and embracing criticism as a valuable gift. An iterative design process led to (1) the creation of the CERTIFIED Manifesto for the ministry, that publicly declares the needed transition mindset, and (2) a practical toolkit to offer policymakers an actionable way to carry out the first steps. The toolkit resembles a low barrier to demystify the desirability of AI application scenarios with a representative group of society. This toolkit contains a Perspectives card deck and a set of 6 canvases. The card deck includes 15 stakeholder cards to empathize with their values, roles, concerns and power. The canvases guide the panel in the steps of Design Thinking and capture the insights from the dialogue. A final toolkit pilot with representative stakeholders of the Quadruple helix showed a cross-fertilization of knowledge and expertise resulting in the ability to identify their blind spots and to formulate a ‘Minimum Viable Policy’ and test set-up to validate their assumptions. This research concludes that if space is provided for proactive policy experimentation, this would give policymakers momentum to reflect on the long-term goals instead of the issues of the day. Thus, offering a way towards becoming an adaptive government.