According to Sundar Pichai, Google trained a new flood forecasting model to predict urban flash floods up to 24 hours in advance, and created Groundsource, an AI methodology using Gemini to identify over 2.6 million historical flash flood events across 150+ countries, which is now open sourced; urban flash flood forecasts are live in Flood Hub to support community safety. As reported by Google via Pichai’s announcement, the combination of Gemini based event extraction and a purpose built forecasting model addresses the data scarcity that has limited city scale flood nowcasting, enabling earlier warnings and operational planning. According to the announcement, enterprises and public agencies can leverage the open dataset for local calibration, model benchmarking, and integration into emergency dispatch, insurance risk models, and municipal resilience planning, while developers can operationalize alerts through Flood Hub outputs.

According to Sam Altman on X, GPT-5.4 is “really good at spreadsheets,” with several finance professionals acknowledging tangible value from the model’s capabilities. As reported by Sam Altman on X, the post highlights improved accuracy and usability in spreadsheet tasks, signaling readiness for workflows like FP&A modeling, sensitivity analysis, and reconciliations. According to the X post, this reaction from finance users suggests rising adoption potential for GPT-5.4 in enterprise finance operations, including automated variance analysis, cash-flow forecasting, and KPI dashboards. For businesses, the opportunity is to pilot GPT-5.4 within governed environments for spreadsheet-heavy processes, integrate it with data warehouses and BI tools, and measure time-to-insight and error-rate reductions.

According to God of Prompt, the Confidence Weighting strategy is a recommended approach for improving the accuracy of numerical estimates, predictions, and forecasts generated by AI models. This method involves requesting each prompt variation to provide a confidence score from 1 to 10, then weighting the answers based on their confidence level—assigning triple weight to high confidence, double to medium, and single to low—before averaging the results. As reported by God of Prompt on Twitter, this strategy allows AI practitioners and businesses to prioritize more reliable outputs, which can enhance decision-making and model trustworthiness in real-world applications.