This working paper evaluates the potential of SatGPT, an AI-powered flood mapping tool that integrates natural language processing, cloud computing, and Earth observation data to support disaster risk management in the Asia-Pacific region. Through capacity development activities in Indonesia and Thailand from 2024 to 2025, participants developed 11 case studies applying SatGPT to real-world scenarios, including mapping exposures, vulnerabilities, and long-term risks, assessing damage and loss, and validating flood data. The results demonstrated the capabilities of SatGPT in automating historical flood analysis, lowering technical barriers, and enabling rapid preliminary risk assessments across different scales. The case studies show that SatGPT can map areas of recurring inundation, highlight potential exposure of key assets and identify vulnerable populations, and provide an initial basis for understanding flood-related impacts. Overall, SatGPT shows strong potential to enhance evidence-based decision-making and support the integration of digital innovations in disaster risk management. It can also generate high-quality, accessible, and standardized risk data for informing investment decisions, guiding priority actions, and making risk more visible to governments, communities, and other stakeholders.