Multi-Agent Pandemic Strategist

Architecture
Application screenshot

Inspiration

The Covid-19 pandemic highlighted the urgent need for fast, informed, and explainable decision-making in public health. We were inspired to build a solution that could simulate and evaluate containment strategies using AI agents—bringing clarity and speed to complex pandemic responses.

What it does

Multi-Agent Pandemic Strategist uses Agentic AI on AWS Bedrock to recommend, critique, and analyze pandemic containment strategies. It delivers real-time, actionable insights with transparent reasoning, helping decision-makers evaluate pros and cons across multiple scenarios.

How we built it

We built the solution using:

AWS Bedrock with the Nova LLM Modular Bedrock Agents:

StrategyAdvisorAgent for strategy recommendations StrategyCritiqueAgent for pros and cons analysis

AWS Lambda & S3 for orchestration and data handling Python & HTML for interface and logic Interactive Query Interface for real-time agent interaction

Agents operate independently or collaboratively to provide layered insights.

Challenges we ran into

Designing agents that balance autonomy with explainability Ensuring contextual relevance across diverse pandemic scenarios Managing multi-agent orchestration without performance bottlenecks Delivering traceable reasoning for each recommendation

Accomplishments that we're proud of

Successfully built a scalable, composable agentic framework Delivered real-time strategy evaluation with transparent logic Demonstrated practical utility in a hackathon setting

What we learned

The value of agent collaboration in complex decision-making Importance of explainability in AI systems How to build scalable architectures using AWS Bedrock Real-world applicability of multi-agent systems in public health

What's next for Multi-Agent Pandemic Strategist

Integrate real-time data feeds for dynamic strategy updates Expand agent capabilities to cover broader health and policy domains Apply the framework to other crisis scenarios (e.g., climate, disaster response) Enhance UI/UX for wider adoption by non-technical stakeholders