Inspiration
Flooding can do more than damage homes and infrastructure. It can contaminate drinking-water sources, exposing entire communities to waterborne diseases. In rural villages across India and Africa, where access to safe water and healthcare may be limited, recognizing these risks early can make a meaningful difference.
Village Water Sentinel was inspired by the idea that AI should do more than analyze problems after they happen. We wanted to build an AI agent that anticipates potential water contamination, communicates clear warnings before people get sick, and helps local health workers prepare an effective response.
Our goal is to turn environmental data into timely, actionable decisions that protect communities and strengthen local resilience.
What it does
Village Water Sentinel is an AI agent that predicts when flooding may contaminate drinking-water sources in rural villages across India and Africa, then coordinates preventive action.
Predicts contamination risk: Pulls live rainfall forecasts and river-level data to assess the likelihood of flooding-related drinking-water contamination.
Scores village-level risk: Assigns each village a risk score from 0–100 and provides a plain-language explanation of the factors contributing to that score.
Automatically alerts families: Sends SMS and voice alerts in local languages, including Hindi, Assamese, Swahili, and Chichewa, advising families to take precautions such as boiling or appropriately chlorinating drinking water.
Coordinates health-worker responses: Generates a visit plan for local health workers, including preparation of chlorine tablets and oral rehydration solution (ORS) packets.
Collects community reports: Allows villagers to text back reports such as SICK or TEST+, helping the agent identify potential cases or positive water-test results.
Escalates emerging risks: If reports indicate possible illness or contamination, the agent escalates the situation to the district level for further attention.
How we built it
Village Water Sentinel combines environmental data, risk assessment, AI-driven decision-making, and automated communication in a coordinated agent workflow.
The system brings together rainfall forecasts and river-level measurements to assess potential flood-related contamination. It translates those signals into a village-level risk score accompanied by an understandable explanation, helping make the results more transparent and actionable.
Based on the assessed risk, the agent can initiate appropriate alert workflows, communicate preventive guidance through SMS and voice channels, and prepare response plans for local health workers.
Community reports provide an additional feedback loop, allowing emerging concerns to be escalated when new information arrives.
Challenges we ran into
One of our main challenges was translating complex environmental signals into a risk assessment that is both useful and understandable. Rainfall and rising river levels can indicate increased danger, but predicting actual drinking-water contamination requires accounting for local conditions and the uncertainty inherent in environmental data.
Another challenge was designing for communities with different languages, communication needs, and levels of infrastructure. An alert is only useful if people can understand it, receive it in time, and act on it.
Accomplishments that we're proud of
We are proud to be building an AI agent around a concrete humanitarian problem rather than limiting AI to information retrieval or passive analysis.
Village Water Sentinel connects several stages of a public-health response:
Environmental signals become village-level risk assessments.
Risk assessments trigger accessible warnings.
Warnings connect to practical preventive actions.
Community reports feed back into the response workflow.
Emerging concerns can be escalated to local authorities.
What we learned
This project taught us that building an effective AI agent requires more than developing a model that produces predictions. The system must interpret changing information, communicate uncertainty, select appropriate actions, and respond when new evidence becomes available.
We also learned the importance of designing technology around the people who will use it. For communities facing water insecurity, an understandable voice alert or a practical health-worker visit plan may be more useful than a sophisticated dashboard.
What's next for Village water sentinel
Our next steps are to improve the accuracy of village-level risk assessments, validate the system against historical flood and contamination data, and test its performance under different environmental conditions.
We also want to expand local-language coverage, strengthen integration with water-quality testing, and refine the way the agent prioritizes communities and coordinates health-worker responses.
In the long term, we envision working with local health workers, public-health organizations, and community partners to pilot Village Water Sentinel in vulnerable communities. Such pilots would help us evaluate whether alerts arrive early enough, whether families can act on the guidance, and whether the system supports more effective local responses.
Log in or sign up for Devpost to join the conversation.