About CityPulse AI Inspiration 💡
Cities are constantly faced with the day to day sorts of problems: potholes, overflowing bins, broken streetlights, water leaks etc. People will invariably raise a complaint about these things but it's often very difficult to categorise, prioritise and track them effectively. That's what inspired us to build CityPulse AI. It's a smart civic complaint management platform that leverages Generative AI and AWS to take the citizen complaint and turn it into insight that can be used to make communities cleaner, safer and smarter.
What We Built 🚀
CityPulse AI enables citizens to submit civic complaints via a simple description (and optionally a photo). Using Amazon Bedrock the system parses the complaint description, determines the category of the complaint, produces a short summary, assigns a priority and suggests the appropriate next step. AWS Lambda is used to process the requests and Amazon DynamoDB is used to store the complaint details and status. There's a web dashboard for administrators to view the complaints and track their progress.
How We Built It 🛠️
We've built the app around AWS serverless services to keep it scalable and manageable. The tech stack is:
Frontend: React (for an interactive, user-friendly interface). AI: Amazon Bedrock for Generative AI powered complaint analysis. Backend: AWS Lambda (for the serverless bit of the business logic). API: Amazon API Gateway (for communication between the front end and back end). Database: Amazon DynamoDB to store the complaints and track their status. Storage: Amazon S3 (for optional complaint image uploads).
Our aim was to build a practical prototype that walks you through the entire process from submitting a complaint right through to AI assisted analysis and status tracking.
Challenges We Faced 🧩
One of our primary challenges was getting the AI responses to be sufficiently consistent and structured for the app to be able to process them. We did this by crafting specific prompts and then validating the output. The integration of the AWS services, the setting of permissions, error handling and keeping the complaint records in sync with the dashboard all needed to be tested.
We also looked at responsible use of AI. The priority levels are recommendations which will require human review rather than being decisions that would automatically be made about public safety.
What We Learned 📚
This project has given us a real-world understanding of how Generative AI can be used in an application via AWS cloud services. We've covered serverless architecture, API integration, database design, prompt engineering, AI output validation and cloud permissions. The key thing we've learned is the importance of having a solid end-to-end workflow in place before you start to add any of the more complex features.
What's Next 🌱
We see CityPulse AI being extended to include multilingual voice reporting, location mapping of complaints, detection of duplicate complaints, real time alerts and analytics which will help identify the recurring problems in the city.
Our Vision 🌍
CityPulse AI is all about connecting the citizen with the civic authorities. The idea is to take the myriad of complaints that citizens have and turn them into a coherent, structured form of information that can be acted upon. We're doing this by leveraging AWS and Generative AI in order to help create more responsive, more transparent and more citizen-centric cities.
Built With
- amazon-api-gateway
- amazon-bedrock
- amazon-dynamodb
- amazon-web-services
- amazon-web-services-(aws)
- artificial-intelligence
- aws-lambda
- generative-ai
- react
Log in or sign up for Devpost to join the conversation.