This is a concept prototype, not a working product

Every number here is illustrative, produced by a toy model for the DAISI Challenge 2026 Round 1. A lot of it does not work yet:

• Genie only answers the four suggested questions. Anything else gets a placeholder reply.
• Nursing homes, Saved briefs and Data and method are not built.
• The forecast range is a stand-in. In the real build it comes from the machine-learning model's uncertainty range, and suggested sites recalculate for it.
• The ML forecast itself is not trained yet. It will be tested against the standard demographic method during the sprint.
• The planning brief is a fixed template, not written by AI.
• Clinic locations on the map are approximate, and areas without HDB survey data are left blank.

The real version would run on Databricks with real data. Please don't quote these figures.