Skip to content
jordan.maulana
ID EN
← Produk
live

Ngaturduit

Telegram-first personal finance — capture transactions in chat, see them everywhere.

Kunjungi situs →
  • Django 5.2
  • DRF
  • Postgres
  • Celery
  • python-telegram-bot
  • React 19
  • OpenAI Agents SDK
  • MCP

A Telegram-first personal finance app for capturing daily money flow, organizing it across wallets, and turning it into reports and AI-assisted conversations.

What it does

  • Capture transactions from Telegram — log income and expenses through a chat bot, no separate app to open.
  • Organize across wallets — group balances by account, with money transfers between wallets recorded as linked transaction pairs.
  • Aggregate automatically — daily and monthly summaries with per-category income/expense breakdowns ready for review.
  • Generate PDF reports — server-rendered reports via WeasyPrint for sharing or archiving.
  • Talk to your data — chat-based conversations powered by the OpenAI Agents SDK, plus an authenticated MCP endpoint so external AI tools can query the same data.

Who it’s for

Telegram-centric users who want friction-free transaction logging — type a message, the bot records it. A web dashboard and a React SPA exist for users who want a richer view of the same data on a larger screen.

How it works

The Telegram bot is the primary capture surface — users record transactions in chat and the bot persists them to a Django backend keyed by telegram_id. The same data is browsable through two parallel surfaces: a server-rendered Django dashboard (with Tailwind templates) and a standalone React SPA that consumes the REST API. AI features layer on top: in-app conversations use the OpenAI Agents SDK, and an authenticated MCP endpoint exposes the same operations to external assistants.

Surfaces

SurfacePurpose
Telegram botPrimary transaction capture
Django dashboardServer-rendered web UI
React SPAStandalone single-page app
REST API v1 / v2Public API for the same data
MCP endpointAuthenticated tool integrations
PDF reportsWeasyPrint-generated reports

Tech summary

  • Backend — Django 5.2 + DRF, PostgreSQL, Celery + Redis, WeasyPrint, WhiteNoise, django-allauth (Google OAuth), Sentry, Langfuse.
  • AI — OpenAI Agents SDK, authenticated MCP via custom view.
  • Botpython-telegram-bot, runs as its own service.
  • Frontend — Vite + React 19 + TypeScript SPA (TanStack Query, Jotai, react-router v7, shadcn/radix, ky); Tailwind v4 for Django templates.
  • Toolinguv (Python 3.12+), pnpm, ruff, djlint.

License

MIT.