Crisis Journalist AI

Dashboard of the application
Create News Section
ARCHITECTURE DIAGRAM

Documents to test the app - link

Inspiration

Crisis Journalist AI was born from the need to combat misinformation during fast-moving global events. In times of crisis, media outlets and journalists are flooded with unverified data, making it nearly impossible to distinguish fact from noise.

Our goal: empower journalists with AI-verified, real-time insights that ensure accuracy, speed, and trustworthiness in reporting.

What it does

Crisis Journalist AI enables news agencies, media outlets, and independent reporters to receive AI-validated, real-time news intelligence.

Using a multi-agent architecture, the system:

Ingests and validates data from multiple verified APIs using a dedicated ingestion agent. Analyzes and summarizes insights with Bedrock-powered reasoning and contextual AI models.
Generates structured news stories ready for publication.
Creates derivative content such as social media posts and short videos, leveraging AWS AI Agent Core (Bedrock) for automated content generation and formatting.

The result — journalists get a verified, concise, and ready-to-publish story in seconds.

🔍 About the Project

Crisis Journalist AI was built to solve a growing issue: information overload and misinformation in crisis situations. Reporters and media outlets struggle to verify data during fast-moving events — our tool uses AI to aggregate, validate, and summarize reliable data in real time.

💡 Inspiration

This project was inspired by the global misinformation crisis seen during major events where quick, accurate reporting is essential. We wanted to create a tool that supports truthful journalism powered by AI-driven data validation.

🧠 What We Learned

We learned how to integrate AI agents, data APIs, and news stream validation pipelines to filter and fact-check large volumes of information. We also deepened our understanding of prompt engineering, data reliability models, and UX for media professionals.

🏗️ How We Built It

Architecture:

Ingestion Layer: AWS Lambda-based API connectors and verification agents for data reliability scoring.
Core AI Layer: Built using AWS Bedrock Agents (multi-agent architecture) for summarization, validation, and content generation.
Frontend: React + GSAP for a modern newsroom dashboard experience.
Backend: FastAPI and DynamoDB,AWS Bedrock for LLM-powered data analysis, orchestrating AI calls and managing data pipelines.
CI/CD: GitHub Actions → AWS Amplify + ECR + Lambda deployments.

Deployment: Hosted on AWS with CI/CD through GitHub Actions. AI Layer: Utilizes Retrieval-Augmented Generation (RAG) to ensure contextually accurate responses.

⚙️ Challenges Faced

Handling data authenticity scoring and ensuring model bias didn’t affect output.

Integrating real-time APIs without losing speed or consistency. Designing a modern, credible interface that resonates with professional journalists.

🧩 Built With

Languages: Python, JavaScript

Frameworks: FastAPI, React, GSAP

Databases: AWS DynamoDB

Cloud Services: AWS Bedrock, S3, Lambda, EC2, Agentcore, AWS Polly, Cloud Front, AWS Amplify

APIs: News API, OpenAI-compatible endpoints

other Tools: MCP

Developmet: Kiro, Strands

🚀 Try It Out

Live Demo: link

GitHub Repository:

Frontend - link

Backend - link

🖼️ Project Media

Alt text

Video Demo:

Project : link

Laptop POV : link

🧾 Additional Info:

App Status: Live

New Amazon Tools Used:

AWS Bedrock (for AI text understanding and summarization)
AWS DynamoDB (for chat history & data validation logs)
AWS Lambda (for event-driven backend execution)

AWS Kiro: We have used AWS Kiro for development and coding for the project.

Agent Coordination: Kiro handled task routing between our 4-agent pipeline — Ingestion, Validation, Summarization, and Media Generation — ensuring synchronized execution.

Execution Monitoring: Provided real-time observability of agent performance and workflow success/failure states.

Scalable Deployment: Simplified running complex, event-driven AI workloads on AWS Lambda through Bedrock integration.

Faster Iteration: Helped us test and deploy new agent logic quickly without manually managing orchestration code.

What's next for Crisis Journalist AI

🧠 Multi-Agent Expansion: Extend our AWS Bedrock agent architecture to include specialized roles — Fact-Checker Agent, _ Trend Analyzer Agent _, and Media Generator Agent — for deeper accuracy and automation.

🌍 Global Coverage: Integrate _ multi-language _ and region-specific news feeds to provide localized, verified insights for crisis events worldwide.

🎥 Media Automation: Evolve from text-based summaries to AI-generated news videos, complete with auto-voicing, captions, and visual data highlights using AWS media services.

📊 Real-Time Crisis Dashboard: Build a live interactive map displaying breaking events, verified sources, and trust-level scoring — powered by continuous AI ingestion.