← Back

AI Chatbot for News Portal — iNews Media Group

MNC Media's four news portals (Okezone, Sindonews, iNews, and IDX Channel), collectively reach millions of Indonesian readers daily. But volume was never the problem. The problem was that readers couldn't navigate it. In early 2026, I led the design and product definition of an AI-powered news assistant that let users ask questions, get summaries, and surface relevant articles through natural language. Embedded directly into each portal, no login required, no page redirect.

Role & services

Lead Product Designer, Product Manager

Timeline

On Progress

Industry / context

News

Problem We Were Solving

 

MNC Media operates four major Indonesian news portals with large daily readership, but users were struggling to extract value from the volume of content available.

  1. Information overload: Readers faced too many articles across too many categories with no intelligent filtering layer
  2. Discovery friction: Finding specific topics, companies, or related stories required manual searching — a slow, low-yield experience
  3. Time constraints: Users wanted quick understanding of complex or developing news without reading multiple full articles
  4. No differentiation: MNC portals lacked a meaningful product innovation that set them apart from major competitors like Kompas, Detik, and Tribun

The business needed a way to increase session depth and position MNC Media as an AI innovation leader in Indonesian digital media.

Goals and Success Metrics


User Goals

  • Find all articles on a topic, company, or person quickly via natural language
  • Understand news context and implications without reading multiple sources
  • Get concise summaries of single or multiple articles
  • Discover trending/breaking news at a glance

Business Goals

  • Drive longer, stickier sessions — convert passive readers into active explorers
  • Differentiate from competitors with a unique AI-native feature
  • Improve content discovery and increase article click-throughs

Research Approach and Baseline

 

Key inputs that shaped the design direction:

  1. Competitor analysis: Benchmarked Kompas, Detik, and Tribun — none offered AI-assisted news Q&A or summarization at the time, confirming a market gap.
  2. Persona research: Defined three primary user segments based on portal analytics and behavioral patterns:
    • Gen Z News Skimmer (18–25, mobile-first, 95%+ mobile, wants instant context)
    • Young Professional Reader (25–40, 60–65% desktop at work, needs depth + efficiency)
    • Young Investor / IDX Reader (20–35, monitors markets, needs company-specific news fast)
  3. Technical landscape review: Evaluated LLM and vector DB options; selected Google Gemini 2.5 Flash + Milvus for cost-efficiency, Bahasa Indonesia support, and GCP ecosystem fit.
  4. Editorial constraints: Aligned with portal teams on content safety requirements — each portal needed an independent content boundary (no cross-portal contamination) and blocklist control.

Ideation and Solution Exploration

 

The core design challenge was scoping an MVP that delivered real value within a strict 6–8 week window, without over-engineering the AI layer.