Name in original language
Estratég.IA Thon - maratona para criação de estratégia de IA em organizações públicas
Initiative overview
Organised by the Conexão Inovação Pública network, the Estratég.IA Thon was an intensive, online marathon positioned at the intersection of strategic foresight and artificial intelligence governance. The programme aimed to democratise access to advanced strategic planning methodologies, specifically focusing on how public institutions can strategically adopt and implement AI. Multidisciplinary teams were provided with a comprehensive "Toolkit" and a curated list of "Prompts" to guide AI models (such as Gemini, ChatGPT, and Claude) as co-pilots in generating AI adoption scenarios, strategic visions, and actionable implementation plans. Over the course of the event, participants learned how to apply rigorous foresight frameworks to design robust AI strategies tailored for public organisations.
Results, outcomes and impacts
Observed results include 17 comprehensive AI strategies (919 pages, 67 projects, 74 minutes of video) tailored for government bodies. Measured via an AI-assisted impact assessment, the intellectual capital generated is valued at approximately BRL 4.25 million, compressing 3–5 months of traditional consulting work into a short agile event. Expected future impacts are substantial: these strategies could directly impact more than 200 million citizens and influence sectors managing approximately BRL 4.5 trillion per year of the national GDP, demonstrating the potential of AI to accelerate public-sector modernisation with a focus on Responsible AI.
Challenges and lessons learned
A primary challenge was ensuring that participants designed specific, viable AI strategic plans rather than generic technological proposals. To mitigate this, the organisers developed and distributed a highly structured "Toolkit" and a specific framework of "Prompts" to guide the human-AI interaction. A key lesson learned was that using generative AI as a co-pilot dramatically enhances the speed and depth of creating AI strategic plans in government, but its effectiveness relies heavily on users being adequately trained in both prompt engineering and foundational strategic planning concepts.



























