Blab AI
OpenAI Academy Partners with Asian Disaster Preparedness Center to Test AI-Driven Flood Response Workflows in Bangkok
In a two‑day event held in Bangkok in June 2026, OpenAI Academy, the Asian Disaster Preparedness Center (ADPC), DataKind, and the Gates Foundation moved a regional AI skills program from theory to prototype testing. Sixty government officials, technical experts, and technology builders from 13 South and Southeast Asian countries worked together to create and evaluate AI‑assisted workflows for flood risk mapping, early warning, and emergency coordination.
The workshop, part of ADPC’s AI‑enabled disaster risk management plan, followed an initial March 2026 session that focused on AI literacy for professionals managing flood and drought risk. The June event, dubbed the AI Skills Jam, shifted to applied development. Participants were grouped into Build, Test, and Learn tracks. In the Build track, disaster‑management practitioners partnered with OpenAI staff and external specialists to identify operational processes that could be supported by AI, set requirements, develop prototypes, and test them with users.
The Learn track covered advanced ChatGPT use, AI‑assisted spreadsheet tasks, and sensitive information management, while the Test track let teams assess prototypes against operational needs. DataKind facilitated feedback on usability, data requirements, institutional fit, and responsible‑use considerations. The organizers did not disclose details about the data used, how information was protected, or whether any tools were connected to live government systems.
Two prototypes emerged from the workshop. A participant from Thailand’s Department of Water Resources built a flash flood risk map using geographic shapefiles and Thai‑language prompts. A participant from Myanmar’s Department of Meteorology and Hydrology created a flood early‑warning dashboard designed to help disaster‑management teams view risk information more quickly. Other sessions explored disaster response and recovery platforms, action planning, historic weather analysis, situation reporting, crisis management, communications, and infographics.
According to Alex Nawar, a member of OpenAI’s Global Affairs team, the basic idea was to pair people who understood operational problems with people who could help turn those problems into working prototypes. Dr. Bhichit Rattakul, Senior Special Advisor at ADPC, emphasized the importance of timely information for communities during the first 24 hours of a disaster. "The quality and speed of information shared with communities can influence response efforts," he said.
ADPC’s Executive Director, Aslam Perwaiz, said the AI Skills Jam is designed to help professionals build practical AI skills, test reusable workflows, and develop solutions that can be applied in day‑to‑day work. He added that the collaborative approach aligns with ADPC’s Strategic Action Plan 2030 for AI‑enabled disaster risk management.
The Gates Foundation, which supported the workshop, highlighted the need for technology to be grounded in local priorities. Pilar Pacheco, Senior Program Officer for Emergency Response, noted that the prototypes may begin as small experiments but could demonstrate how technology can help responders manage information more effectively and strengthen service delivery.
Selected workflows and prototypes will undergo further development. The partners plan to refine the strongest outputs and may turn them into Digital Public Good‑oriented packages that other disaster‑management organizations can reuse. However, no publication timetable or selection criteria have been announced, and the partners have not disclosed which projects will progress or when materials will be released.
At this stage, the prototypes remain workshop outputs rather than operational systems. No accuracy results, user‑testing figures, deployment dates, or evidence of use during an emergency were included in the announcement. The next steps will determine whether the AI‑assisted tools can be integrated into national disaster‑management workflows and how they will be governed to ensure responsible use.
The event illustrates a growing trend of regional collaboration to apply generative AI to disaster‑risk reduction, but it also underscores the need for clear pathways from prototype to deployment, robust data governance, and ongoing evaluation of effectiveness in real‑world settings.
The workshop, part of ADPC’s AI‑enabled disaster risk management plan, followed an initial March 2026 session that focused on AI literacy for professionals managing flood and drought risk. The June event, dubbed the AI Skills Jam, shifted to applied development. Participants were grouped into Build, Test, and Learn tracks. In the Build track, disaster‑management practitioners partnered with OpenAI staff and external specialists to identify operational processes that could be supported by AI, set requirements, develop prototypes, and test them with users.
The Learn track covered advanced ChatGPT use, AI‑assisted spreadsheet tasks, and sensitive information management, while the Test track let teams assess prototypes against operational needs. DataKind facilitated feedback on usability, data requirements, institutional fit, and responsible‑use considerations. The organizers did not disclose details about the data used, how information was protected, or whether any tools were connected to live government systems.
Two prototypes emerged from the workshop. A participant from Thailand’s Department of Water Resources built a flash flood risk map using geographic shapefiles and Thai‑language prompts. A participant from Myanmar’s Department of Meteorology and Hydrology created a flood early‑warning dashboard designed to help disaster‑management teams view risk information more quickly. Other sessions explored disaster response and recovery platforms, action planning, historic weather analysis, situation reporting, crisis management, communications, and infographics.
According to Alex Nawar, a member of OpenAI’s Global Affairs team, the basic idea was to pair people who understood operational problems with people who could help turn those problems into working prototypes. Dr. Bhichit Rattakul, Senior Special Advisor at ADPC, emphasized the importance of timely information for communities during the first 24 hours of a disaster. "The quality and speed of information shared with communities can influence response efforts," he said.
ADPC’s Executive Director, Aslam Perwaiz, said the AI Skills Jam is designed to help professionals build practical AI skills, test reusable workflows, and develop solutions that can be applied in day‑to‑day work. He added that the collaborative approach aligns with ADPC’s Strategic Action Plan 2030 for AI‑enabled disaster risk management.
The Gates Foundation, which supported the workshop, highlighted the need for technology to be grounded in local priorities. Pilar Pacheco, Senior Program Officer for Emergency Response, noted that the prototypes may begin as small experiments but could demonstrate how technology can help responders manage information more effectively and strengthen service delivery.
Selected workflows and prototypes will undergo further development. The partners plan to refine the strongest outputs and may turn them into Digital Public Good‑oriented packages that other disaster‑management organizations can reuse. However, no publication timetable or selection criteria have been announced, and the partners have not disclosed which projects will progress or when materials will be released.
At this stage, the prototypes remain workshop outputs rather than operational systems. No accuracy results, user‑testing figures, deployment dates, or evidence of use during an emergency were included in the announcement. The next steps will determine whether the AI‑assisted tools can be integrated into national disaster‑management workflows and how they will be governed to ensure responsible use.
The event illustrates a growing trend of regional collaboration to apply generative AI to disaster‑risk reduction, but it also underscores the need for clear pathways from prototype to deployment, robust data governance, and ongoing evaluation of effectiveness in real‑world settings.
