BreatheWise AI

Emergency button
General Advisory
Full dashboard view

BreatheWise AI – An AI-powered Respiratory Health Companion

Our Story

Every winter, Delhi's air quality reaches hazardous levels, exposing millions of people to severe health risks. However, not everyone is affected equally. Senior citizens, asthma patients, individuals with COPD, and those suffering from chronic respiratory illnesses often face the greatest danger. While numerous applications display Air Quality Index (AQI) values and pollution maps, they leave users with an unanswered question:

"Given my health condition, what should I do right now?"

This question inspired us to build BreatheWise AI.

Rather than creating another AQI dashboard, we wanted to build an **AI-powered respiratory health companion that transforms complex environmental data into personalized, actionable healthcare guidance. Our goal was to move beyond simply informing users about pollution levels and instead help them make safer daily decisions.

To achieve this, we designed BreatheWise AI as an agent-based intelligent assistant. Users begin by entering basic information such as their age, respiratory condition, current symptoms, and location. Simultaneously, the system retrieves current air quality information and combines it with trusted public health knowledge.

At the heart of our solution is a Retrieval-Augmented Generation (RAG) pipeline powered by Elastic Cloud. Instead of relying solely on a language model, BreatheWise AI first retrieves relevant respiratory health guidelines, WHO recommendations, and pollution advisories from Elastic. These verified documents are then supplied to the AI model, enabling it to generate recommendations that are both contextual and grounded in trusted medical guidance.

The AI performs three major tasks. First, it assesses the user's respiratory risk based on their health profile and the surrounding air quality. Second, it generates personalized preventive recommendations such as avoiding outdoor activity, wearing an N95 mask, or postponing strenuous exercise. Finally, if the user's risk is determined to be high, the healthcare navigation component recommends nearby hospitals or pulmonologists, providing users with practical next steps instead of simply presenting data.

One of our biggest challenges was balancing personalization with responsible AI. We wanted the system to provide meaningful health guidance without making medical diagnoses. This required carefully structuring prompts, grounding every recommendation in retrieved medical knowledge, and ensuring the AI acted as a decision-support companion rather than a replacement for healthcare professionals.

Another challenge was designing an application that remained true to the hackathon theme while differentiating itself from traditional AQI monitoring platforms. Instead of focusing on pollution visualization, we shifted our perspective toward human health outcomes , allowing every feature to revolve around helping vulnerable communities breathe safer.

Through this project, we explored how agentic AI systems, semantic search, and Retrieval-Augmented Generation can work together to solve real public health problems. More importantly, we learned that the true value of AI lies not in generating impressive responses but in helping people make informed, trustworthy decisions.

BreatheWise AI is only the beginning. In the future, we envision integrating wearable devices such as pulse oximeters and smart inhalers, providing hyper-local exposure predictions, delivering proactive health alerts during pollution spikes, supporting multilingual accessibility, and collaborating with healthcare providers for evidence-based respiratory care. Ultimately, we hope to evolve BreatheWise AI into a comprehensive digital health companion that empowers vulnerable communities to navigate Delhi's air pollution with confidence.