{
  "id": 3971728,
  "title": "Building Mathondu: How We Built an Autonomous AI Merchant & Store Factory with Google ADK, Gemini 3.5, and Cloud Run",
  "url": "https://urgent.news/2026/08/28/building-mathondu-how-we-built-an-autonomous-ai-merchant-store",
  "topic": "ai",
  "section": "AI",
  "published": "2026-08-28T13:41:51.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/navaneethalgorithm/building-mathondu-how-we-built-an-autonomous-ai-merchant-store-factory-with-google-adk-gemini-43oc"
  },
  "original_language": "en",
  "account": "The article describes the development of Mathondu, an autonomous AI merchant and store factory platform built using Google ADK, Gemini 3.5, and Cloud Run. The platform aims to eliminate the friction faced by small commerce businesses in India who rely on WhatsApp and UPI for daily operations.\n\nThe system architecture consists of a fleet of specialized sub-agents coordinated by the onboard_orchestrator, a top-level ADK agent. These sub-agents include store_extract_agent, product_data_agent, missing_fields_agent, and catalog_image_agent, each responsible for specific tasks such as extracting business information, converting product information, identifying missing data, and generating catalog images.\n\nWhen a merchant provides product images, the pipeline processes the information through these specialized agents. The store_extract_agent uses Gemini 3.5 Flash Lite to extract important business details like business name, operating hours, delivery radius, and store metadata. The product_data_agent converts messy product information into structured schemas, including product names, categories, prices, and variant options. The missing_fields_agent identifies and generates targeted follow-up questions for any missing critical information.\n\nFor merchants lacking product images, the catalog_image_agent generates high-resolution catalog photography using Gemini 3.1 Flash Lite. This allows merchants to create professional-looking catalogs without the need for a photographer or manual image editing.\n\nThe platform generates real deliverables such as Progressive Web App (PWA) storefronts, branded PDF catalogs, custom Android APKs, and WhatsApp ordering experiences. Merchants can request an Android APK, which triggers a build job in Firestore, a Cloud Run container, Bubblewrap CLI compilation, Android binary signing, Firebase Storage upload, and a secure download link with an install QR code.\n\nAdditionally, the platform supports WhatsApp commerce, enabling customers to interact with merchants directly through the platform. WhatsApp orders are processed through an interactive shopping experience, maintaining cart and conversation state in Cloud Firestore. Real-time order events alert sellers with direct links to the admin console.\n\nOverall, Mathondu automates and streamlines the process of creating digital stores for small merchants, providing them with a complete commerce solution without the need for technical expertise.",
  "summary": "Disclosure: I created this piece of content for the purposes of entering the Google All Things Agentic Hackathon on Devpost. 🌟 The Problem: Why Small Commerce Needs Autonomous Agents, Not Just Chatbots Millions of small merchants, home bakers, and neighbourhood store owners across India run their daily business through WhatsApp and UPI . While messaging works for basic conversations, scaling a…",
  "key_points": [
    "Mathondu platform automates AI merchant creation for small businesses",
    "Built using Google ADK, Gemini 3.5, and Cloud Run technology stack",
    "Eliminates friction for WhatsApp and UPI-dependent Indian commerce"
  ],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}