Wargio — AI Agent for Indonesia's Micro-Retailers

Built for Google Cloud Rapid Agent Hackathon 2026 | MongoDB Track

What is Wargio?

Wargio is an AI business assistant for Indonesian warung owners. Ask in natural Bahasa Indonesia or English — stock, debts, sales — and the agent reasons over live MongoDB Atlas data, then takes action (not just answers).

Demo

Resource	Link
Live app	https://wargio.adindamochamad.com
Demo video	(YouTube — see Hari 7 script)
Judge verify	export WARGIO_PRODUCTION_URL=https://wargio.adindamochamad.com && bash scripts/judge_verify.sh

Toggle EN in the header for English UI + agent responses (for judges).

Architecture

flowchart LR
  U[User ID/EN] --> FE[Next.js UI]
  FE --> API[FastAPI VPS]
  API --> CLS[Gemini 2.5 Flash\nintent classification]
  API --> REG[Regex fallback\nwrite intents]
  API --> HND[Intent handlers]
  HND --> ATL[(MongoDB Atlas)]
  HND --> MCP[MCP-equivalent layer\nfind aggregate insert update]
  MCP --> ATL
  API -.-> ADK[Agent Engine ADK\ndeployed]
  CLS -.-> GCP[Google Cloud]

Loading

Production path: User → Next.js → FastAPI → Gemini classification (+ regex for write) → atlas_tools (same operations as MongoDB MCP) → Atlas.

Why PyMongo-equivalent in production? MCP stdio adds latency; production uses identical tool semantics (find, aggregate, insertOne, updateOne) via atlas_tools. MCP server verified locally: npm run verifikasi:mcp.

Tech Stack

Layer	Technology	Version / notes
Agent classification	Gemini 2.5 Flash	Google AI / Vertex
Agent Engine	Google ADK	AGENT_ENGINE_ID deployed
Data	MongoDB Atlas M0	Live, wargio_demo
MCP	mongodb-mcp-server	Verified + optional live stdio
Vector search	Atlas Search 768d cosine	products_vector_index
API	Python 3.11+, FastAPI, Pydantic v2
DB driver	PyMongo Async
Frontend	Next.js App Router, TypeScript, Tailwind
Deploy	Docker + Nginx + HTTPS (VPS)	docs/deploy-vps.md

MongoDB Integration

• Collections: products, transactions, customers, agent_sessions
• 8 intents: check_stock, check_debt, restock_alert, sales_report, record_sale, record_payment, debt_collection, sales_forecast
• Fuzzy products: exact alias → partial (≥0.8) → vector search (≥0.85, Gemini embedding 768d)
• Write safety: user confirmation required before insertOne / updateOne
• Aggregations: sales reports, day-of-week forecast

Google Cloud Integration

Service	Usage
Gemini 2.5 Flash	Intent classification; product embeddings (768d)
Agent Engine (ADK)	Deployed agent (docs/setup-agent-builder.md)
Vertex AI	Optional path when GOOGLE_CLOUD_PROJECT set

Check production: curl -s https://wargio.adindamochamad.com/api/health

Local Development

Prerequisites

• Python 3.11+
• Node.js 20+
• MongoDB Atlas cluster + connection string
• Google Cloud project + Gemini API key (optional but recommended)

Backend

cd backend
python -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt

cp ../.env.example ../.env   # MONGODB_URI, GEMINI_API_KEY, etc.

python ../scripts/buat_indeks.py
python ../scripts/buat_vector_index.py --bebaskan-slot-sample
python ../scripts/seed_data.py
python ../scripts/isi_embedding_produk.py

uvicorn app.main:aplikasi --reload --host 0.0.0.0 --port 8000

Frontend

bash scripts/jalankan_dev.sh          # API + MCP off = UI cepat
cd frontend && npm install && npm run dev

Open http://localhost:3000

Verify MCP (hackathon requirement)

npm run verifikasi:mcp          # MCP tools → Atlas
npm run verifikasi:mcp-live     # stdio pool live (lambat)

Production

bash scripts/siapkan_env_production.sh https://your-domain.com
docker compose -f deploy/docker/docker-compose.yml --env-file deploy/.env.production up -d --build
bash scripts/judge_verify.sh

Full guide: docs/deploy-vps.md

Environment Variables

See .env.example. Never commit secrets.

Variable	Purpose
MONGODB_URI	Atlas connection
GEMINI_API_KEY	Classification + embeddings
AGENT_ENGINE_ID	ADK deployed agent
MCP_LIVE_ENABLED	false in production (latency)
API_INTERNAL_URL	SSR dashboard fetch (Docker: http://api:8000)

Running Tests

# Backend (needs MONGODB_URI for integration tests)
cd backend && pip install -r requirements-dev.txt
pytest ../tests -q

# Frontend
cd frontend && npm run test

# Hari 6 gates
python scripts/verifikasi_hari6.py

# Production smoke (8 intent + write + EN)
export WARGIO_PRODUCTION_URL=https://wargio.adindamochamad.com
bash scripts/smoke_production.sh

Submission

• Devpost copy: docs/devpost-submission.md
• Demo video script: docs/hari7-demo-video.md
• Track: MongoDB

License

MIT — see LICENSE.