AI-First Global News Insight (Newspapler) — Project Report

1. Project Overview

Newspapler is an innovative global online newspaper platform designed to operate through collaboration between an AI Reporter system and a single human editor/operator.

The platform adopts a clean and minimalist editorial design inspired by The New York Times, featuring serif typography, a structured grid-based layout, and subtle minimalist visual elements. The project is designed to deliver news content simultaneously in Korean and English, with the long-term goal of building a multilingual global news platform.

The core concept is to position AI not merely as a content-generation tool, but as a specialized AI journalist with a distinct professional persona and editorial voice.

2. Initial Site Development Planning

The central concept of the initial planning process was to use AI as a professional journalist rather than simply as an auxiliary writing tool.

Based on the original guide.md and ask.md specifications, the initial product concept included the following core systems and features.

AI Reporter Registration System (Persona)

Administrators can create and manage AI reporters with different areas of expertise.

For example:

• A reporter specializing in history

• A reporter specializing in media criticism

• A reporter specializing in international affairs

• A reporter specializing in technology or culture

Each AI reporter is defined by a set of professional characteristics, including:

• Academic degree or educational background

• Area of expertise

• Professional strengths

• Writing style and tone

• Editorial guidelines and preferences

These attributes are stored in the database as part of the reporter's prompt configuration. When an article is generated, the selected AI reporter uses its predefined persona to produce content consistent with its area of expertise and editorial voice.

Automated Article Generation Workflow (Core Engine)

The core engine is designed around a queue-based automated journalism workflow.

1. The administrator creates an article assignment containing the initial topic or draft, background URLs or reference materials, and the target section.

2. The assignment is added to the article generation queue.

3. The selected AI reporter analyzes the provided sources, performs scraping and summarization where applicable, and develops the necessary context.

4. The AI generates both Korean and English versions of the article.

5. The generated article is stored in a pending state for editorial review.

6. The human editor reviews the generated content, determines its placement on the website, and approves or publishes the article.

This workflow is intended to establish a human-in-the-loop editorial system, in which AI handles research and initial content production while the human operator retains final editorial control.

UI/UX Design and Monetization Strategy

The visual design follows a refined editorial aesthetic inspired by traditional newspaper layouts.

Key design elements include:

• A newspaper-style header and logo

• Serif-based typography

• Prominent date and publication information

• Responsive Pinterest-style masonry article grids

• Minimal borders and visual separators

• Clear editorial hierarchy between headlines, summaries, and article content

PWA and Mobile Experience

The platform is also designed with a mobile-first and Progressive Web App (PWA) experience in mind.

Planned features include:

• Infinite scrolling

• App-like bottom navigation

• Mobile-optimized article layouts

• Push notifications for breaking news

• Responsive layouts across desktop, tablet, and mobile devices

Monetization Strategy

The initial monetization plan focuses primarily on digital advertising, including:

• Native in-feed advertisements

• Context-aware advertisements within article content

• Fixed sidebar advertisements

• Other strategically placed advertising units designed to maintain the overall editorial aesthetic

The goal is to integrate advertising into the user experience without significantly disrupting the newspaper-style presentation.

Editorial and Emotional Features

In addition to news content, the platform includes more atmospheric and editorially distinctive features.

One example is the “Poem of the Day” modal, designed to introduce an elegant and emotional element to the overall reading experience.

This feature is intended to give Newspapler a more distinctive editorial identity beyond conventional automated news aggregation.

3. Technology Stack

Newspapler combines modern web technologies, a serverless backend architecture, and generative AI APIs.

Frontend

• Next.js 16.1.6 — App Router

• React 19

• TypeScript

Styling

• Tailwind CSS v4

• clsx

• tailwind-merge

Animation and Icons

• Framer Motion

• Lucide React

Rich Text Editing

• Tiptap — Headless rich-text editor

Backend and Database

• Supabase

• PostgreSQL

• @supabase/ssr

• @supabase/supabase-js

Supabase serves as the primary backend-as-a-service platform, providing database management, authentication-related infrastructure, and server-side integration.

AI Integration

• Google Generative AI

• Gemini API

• @google/generative-ai

Gemini is used as the core generative AI engine for tasks such as source analysis, summarization, contextual understanding, and multilingual article generation.

Utilities and Security

• date-fns — Date and time handling

• DOMPurify — XSS protection and HTML sanitization

4. Database Architecture

Newspapler uses Supabase PostgreSQL as its primary database. The database is designed around a set of interconnected entities that support the AI journalism workflow, editorial management, content organization, and website configuration.

articles — Articles

Stores the core article data, including:

• Korean and English titles

• Korean and English summaries

• Article body content

• Publication status

• Creation and publication timestamps

• Author information

• Associated category

• AI-related metadata

The articles table serves as the central content entity of the platform.

categories — Categories

Defines the site's editorial classification system.

Key fields include:

• Korean category name

• English category name

• Slug

• Display order

• Layout configuration

This structure allows the editorial team to organize and present articles consistently across different sections of the website.

ai_reporters — AI Reporters

Stores the personas and professional characteristics of AI reporters.

Key fields include:

• Nickname

• Area of specialization

• Academic degree

• Writing style

• Persona and editorial instructions

These settings are used to customize the AI's behavior during article generation.

article_generation_queue — Article Generation Queue

Manages article-generation requests submitted by administrators.

The table contains information such as:

• Initial article topic or draft

• Background URLs

• Assigned AI reporter (reporter_id)

• Target category (category_id)

• Processing status

• Reference to the resulting article (result_article_id)

This table functions as the central queue for the automated content-generation pipeline.

homepage_slots — Homepage Slots

Controls how articles are positioned on the homepage.

The system can assign articles to specific editorial positions, such as:

• Main headline

• Featured article

• Recommended article

• Sidebar content

Each slot is associated with an article and contains information necessary to control its homepage placement.

site_settings — Site Settings

Provides centralized control over global website configuration through a single configuration record.

Settings include:

• Multilingual site name

• SEO and metadata configuration

• Google AdSense settings

• Logo configuration

• Font settings

• Other global website options

Centralizing these settings allows the administrator to manage the site's overall identity and configuration from a single location.

Additional Supporting Tables

The platform also includes several supporting entities:

• poems — Manages content for the “Poem of the Day” feature.

• inquiries — Stores customer and user inquiries.

• subscribers — Stores email addresses for newsletter and subscription services.

• pages — Manages static pages such as About Us, Terms of Service, and other informational content.

Together, these database entities provide the foundation for Newspapler's AI-driven newsroom, editorial workflow, multilingual publishing system, homepage management, and supporting website functions.