Track Telegram expenses with GPT-4 and Google Sheets (self-learning categories)

💸 HOW IT WORKS — AI TELEGRAM EXPENSE TRACKER

This workflow transforms natural Telegram messages into structured expenses using AI — without forms, manual typing, or complex inputs.

Simply send a message like:

Groceries 23€ yesterday

The workflow validates the sender, understands the intent, extracts structured data, and prepares the expense for approval before saving.

────────────────

🔄 WORKFLOW OVERVIEW

🟩 1. Secure Input Layer
Incoming Telegram messages are checked against a list of approved Chat IDs to ensure only authorized users can create expenses.

🟦 2. AI Expense Detection
An AI layer analyzes the message and decides whether it represents a real financial transaction.
Non-expense messages are safely ignored to avoid noise in your data.

🟨 3. Smart Category Intelligence
Existing categories are loaded from Google Sheets and compared with the message content.
If no suitable category exists, the workflow can suggest and learn new categories over time.

🟪 4. Structured Data Extraction
AI converts natural language into structured fields:

date

amount

category

description

shared vs personal expense

Supports German and English input.

🟥 5. Human Approval & Storage
Before saving, the user confirms the extracted result directly via Telegram.
After approval, the expense is appended to Google Sheets automatically.

────────────────

📋 SETUP REQUIREMENTS

Before using this workflow, make sure the following components are ready:

1️⃣Telegram Bot
Create a Telegram bot using BotFather and connect it to the Telegram Trigger node in n8n. Detailed setup instructions can be found here.

2️⃣LLM API Access
An API Key for a Large Language Model (LLM) is required for:

expense detection

category matching

structured data extraction

Add your API credentials inside the AI node configuration.

3️⃣Google Sheets

Create two Google Sheets before importing the workflow.

EXPENSES
Required columns:
date, amount, category, description, common_expense, Person

EXPENSE_CATEGORIES
Required columns:
category, description, examples

The workflow reads existing data and appends new entries automatically.

────────────────

💡KEY FEATURES

• AI-powered expense detection from natural language
• Self-learning category system
• Human-in-the-loop approval step
• Multi-language support (DE & EN)
• Clean Google Sheets integration
• Designed for real-life shared finance tracking

────────────────

👥MULTI-USER SUPPORT

Built for couples, roommates, or teams.

Add multiple Chat IDs in:
Security — Allow Approved Chat IDs

Each expense is automatically tagged with the sender.
Shared expenses are stored as true in the common_expense column, while personal expenses default to false unless shared spending is detected.

This allows easy downstream analysis, dashboards, or automation.