kushall-j-4/sagar-ai-poc

Purpose and Runtime

SAGAR is a static, client-side Next.js demo illustrating conversion of government webpage content into Schema.org/JSON-LD-style semantic data and the resulting improvement to AI answers. It has no real ingestion, AI, database, or backend service; all displayed processing and graph data are static/demo state.

Runtime: Next.js 14 App Router, React, TypeScript, Tailwind CSS, and Lucide icons. Production output is a static export intended for GitHub Pages.

Application Entry and Shell

The App Router entry point is app/page.tsx, which owns the guided multi-screen demo flow and likely switches between the input, comparison, AI-answer, and processing views. app/layout.tsx provides the shared document/layout boundary, while app/globals.css establishes global visual styling.

Paths:

• app/page.tsx

• app/layout.tsx

• app/globals.css

Stage 1: Government URL Intake

The initial UI captures a government-site URL as the conceptual source material for semantic processing. This is a presentation/input boundary rather than a network integration: the URL advances the demo but does not invoke a scraping or processing backend.

Paths:

• app/page.tsx

• app/components/SourceView.tsx

Stage 2: Semantic Schema Transformation

The schema-comparison view presents the central product claim: unstructured/current source representation is transformed into an AI-readable semantic structure based on standards such as Schema.org and JSON-LD. This is visualized client-side from predefined data, not generated dynamically.

Paths:

• app/components/SchemaComparisonView.tsx

• app/page.tsx

Stage 3: Citizen/AI Answer Comparison

The citizen-question stage contrasts AI results derived from ordinary webpage content with results supported by structured semantic data. It represents the consumer-facing outcome of the transformation pipeline and remains isolated from external LLM APIs.

Paths:

• app/components/CitizenQuestionView.tsx

• app/page.tsx

Stage 4: Processing Pipeline and Knowledge Graph

The backend-process view visualizes the conceptual internal pipeline—source processing, semantic extraction/structuring, and knowledge-graph representation. Interactive graph nodes are a frontend visualization boundary; there is no persisted graph database, processing worker, or API behind it.

Paths:

• app/components/BackendProcessView.tsx

• app/page.tsx

Deployment Boundary

The application is configured as a static Next.js export, so the build produces HTML/CSS/JS deployable without a Node server. GitHub Actions handles deployment to GitHub Pages on repository pushes; deployment-specific behavior is documented separately.

Paths:

• .github/workflows/deploy.yml

• next.config.js

• DEPLOYMENT.md