Project-9: Agent Persona & Capability Profile

1. Problem Statement

BAN solves the Information Asymmetry in Bengaluru's civic services. Citizens have problems, the government has data, but there is no "Reasoning Tunnel" to connect the two without human bureaucratic overhead.

2. Solution Overview

The agent acts as a Civic Middleware. It ingests a user's problem, executes the SIFT (Search, Identify, Filter, Transform) loop across civic databases, and provides a verified, cited solution.

3. Key Features

• Persona-Driven Logic: Switch between a Planner, Activist, or Architect.
• Geospatial Triage: Real-time markers for "Incident Zones."
• Evidence Verification: Thumbs up/down feedback loop for model fine-tuning.
• MCP Tool Extraction: Uses structured functions to pull current urban data.

4. Technologies Used

• Google Gemini: The core neural reasoning engine.
• Elasticsearch: The librarian of city data.
• AWS Lambda: Serverless execution of civic tools.
• Pigeon Maps: Real-time geospatial visualization.

5. Target Users

• City Officials: For rapid triage of incoming ward issues.
• Civic Tech Builders: As a base layer for urban apps.
• Bengaluru Citizens: For reliable, evidence-backed urban navigation.

Built With

• css
• geminiapi
• html
• python
• react
• typescript