Forensiq

Main Dashboard
Suspect Database Search
System Manual
Team Members

Inspiration

The inspiration for Forensiq comes from the critical need for precision in the Indian legal system. With the transition to the BNS (Bharatiya Nyaya Sanhita) and BNSS (Bharatiya Nagarik Suraksha Sanhita) framework, legal professionals face an enormous challenge in cross-referencing massive amounts of legacy documentation with new legal codes. We wanted to build a "Reasoning Engine" that acts as a second pair of eyes for investigators, ensuring that no discrepancy between an FIR (First Information Report) and witness testimony goes unnoticed.

What it does

Forensiq is a high-end forensic dashboard that automates the detection of inconsistencies in legal evidence.

FIR Repository: Allows users to upload and digitize legal case records.

Multimodal Integration: Ingests video testimonies and generates live transcriptions.

AI Analysis: Uses the Gemini 3 Pro model to analyze evidence and highlight discrepancies in a dedicated "AI Thoughts" sidebar.

Legal Mapping: Operates within the modern Indian law framework to provide high-accuracy forensic insights.

How we built it

We utilized a modern, full-stack approach to ensure both performance and security:

Backend: Built with FastAPI in Python to handle AI logic and secure data processing.

Frontend: Developed using React, Tailwind CSS, and Lucide-React for a professional, dark-themed forensic interface.

AI Engine: Integrated the Google GenAI SDK to leverage the reasoning capabilities of Gemini 3 Pro.

Environment: Configured with python-dotenv to securely manage sensitive API credentials.

Challenges we ran into

The development process was not without its hurdles. We encountered several technical "roadblocks":

Environment Setup: We faced initial "command not found" and "Import could not be resolved" errors while setting up the FastAPI environment in VS Code.

CORS Management: Bridging the gap between our frontend (Vite/Live Server) and the backend required careful configuration of Cross-Origin Resource Sharing.

API Key Integration: We hit a ValueError regarding "Missing key inputs" because our backend was initially unable to find the GEMINI_API_KEY in the local environment.

Accomplishments that we're proud of

Truth Architecture: Successfully building a system that can take raw text from an FIR and a video transcript and identify logical contradictions between them.

Clean UI: Designing a specialized forensic interface that feels intuitive for legal professionals while maintaining a high-tech "lab" aesthetic.

Real-time Sync: Implementing live transcription status and status tracking for the AI analysis process.

What we learned

Through this project, we deepened our understanding of Multimodal AI—specifically how to use Large Language Models (LLMs) as reasoning engines rather than just text generators. We also learned the importance of robust environment configuration and how to manage the complex data flow between a React frontend and a Python backend in a local development setting.