Inspiration
Navigating complex electoral data, legislative assembly regulations (such as majority thresholds in assemblies like a 234-seat house), and voter turnout metrics often involves manual friction and scattered information. We wanted to build an intelligent platform that brings automated clarity and precision to civic data.
What it does
CivicPulse AI is an autonomous, heavy-duty agentic platform powered by Gemini 3.5 Flash. Instead of a basic chatbot, it features native function calling and tool-use execution. When users enter queries regarding legislative majorities or electoral turnout statistics, the agent autonomously invokes custom Python code tools to compute precise mathematical data and displays a real-time execution trace log.
How we built it
- Backend & Agent Engine: Built with Python, Flask, and the official Google GenAI SDK.
- Autonomous Tool-Use: Integrated custom Python functions for real-time mathematical computations and electoral analysis.
- Frontend: Engineered a modern enterprise-grade command-center UI with dark-mode styling and live telemetry logs.
- Deployment Ready: Fully containerized using Docker for seamless cloud scaling.
Accomplishments that we're proud of
- Successfully implementing native function calling so the AI dynamically executes code instead of just guessing numbers.
- Building a complete, production-grade agentic system from scratch with live telemetry and clean UI design.
What we learned
We mastered advanced agentic design patterns, function-calling schemas, and integrating backend Python logic seamlessly with state-of-the-art LLM pipelines.
Log in or sign up for Devpost to join the conversation.