Inspiration Google Maps, and public-facing data tools like CalEnviroScreen and the Healthy Places Index, which show the power of visualizing complex data to empower communities.
What it does This project is a specialized GeoAI analyst designed to serve a "Health & Environmental Equity" map. It answers complex, location-specific questions from a diverse range of users (from residents to policymakers). It does this by interpreting a DATA_PACKET and providing clear, ethical, and accessible data-driven insights.
How we built it The system is built on a powerful large language model (Gemini 2.5). The model is guided by a highly detailed "Master System Prompt" that defines its core ethical directives, its analytical process, and its "Persona-Adaptive Logic." This logic allows it to change its tone and technical depth depending on the user. This AI is paired with a data analysis component (using Python) that processes the raw CSV data to extract key statistics, which are then fed to the AI for interpretation.
Challenges we ran into Ethical Analysis: A major challenge is processing sensitive data (like poverty, asthma rates, and pollution) without creating negative or judgmental descriptions of a community. The system had to be explicitly programmed to "Avoid Stagmatization" and present facts objectively.
Avoiding Harmful Advice: We had to ensure the AI never gives absolute personal recommendations (e.g., "don't buy a house here"). The challenge was teaching it to instead summarize objective data factors relevant to the user's decision.
Translating Jargon: Translating technical metrics (like "PM2.5" or "MORT Group Measure Count") into simple, clear, and accurate explanations for non-expert users was a significant challenge that required building a detailed reference guide.
Accomplishments that we're proud of Persona-Adaptive Logic: We are proud of the system's ability to tailor its response to the user's likely persona (e.g., Community Resident, Health Worker, or City Official). This makes the complex data far more accessible and useful to everyone.
Data-Driven Ethical Framework: Successfully integrating a strong set of "Core Ethical Directives" directly into the AI's logic, ensuring its analysis is objective, data-driven, and non-judgmental.
Clarity and Context: The system doesn't just provide numbers; it explains what they mean on a recognized scale (e.g., comparing a PM2.5 level to WHO guidelines), which provides crucial context.
What we learned We learned that when dealing with sensitive health and demographic data, the AI's "personality" and ethical rules are just as important as its ability to analyze data. A simple data-dump is not helpful and can even be harmful. The key is to create a "translator" that turns complex data into clear, objective, and actionable insights.
What's next for LAccess Create more layers for other resources
Build a better UI with React.js
Partner with non-profit organizations to help them
Let users upload their own data
Integrate a LLM to answer questions on the data
Log in or sign up for Devpost to join the conversation.