n8n Bank SMS to Google Sheets + Telegram Claude Bot

Bank SMS → Google Sheets ledger + Claude-powered Telegram answers (n8n workflow)

Turn forwarded bank SMS alerts into structured expense records in Google Sheets, and ask “How much did I spend on X?” directly in Telegram—powered by an Anthropic Claude LangChain agent that reads your ledger and review queue.

What this workflow does

• Receives bank SMS via webhook: An authenticated secured webhook accepts forwarded SMS messages.
• Parses and normalizes transaction text: The workflow extracts fields such as amount, debit/credit, date, merchant, payment method, reference ID, and account last 4, while rejecting OTPs and non-allowlisted senders.
• Routes complete vs incomplete messages:
  • Complete, valid transactions are upserted into the Expenses tab and return a 200 JSON confirmation.
  • Incomplete but not rejected SMS messages are upserted into Review_Queue and return a 200 JSON response indicating manual review is needed.
  • Rejected SMS (e.g., OTP or unknown sender) returns a 422 JSON rejection response.
• Answers questions in Telegram using Claude: On incoming Telegram messages, the workflow sends your question to a LangChain agent using Anthropic Claude with short-term conversation memory. The agent reads transactions and the Review_Queue from Google Sheets, uses a calculator tool for exact math, and replies with a plain-text answer in the same chat.

Use cases

• Automate personal or business expense tracking from bank SMS without manual entry.
• Catch ambiguous SMS alerts in Review_Queue instead of losing transaction data.
• Ask spending questions in Telegram while trusting the ledger and exact calculations.

Technical details

• n8n nodes/tools: webhook, if, code, telegram, and Google Sheets (plus sticky note for documentation).
• Google Sheets setup: Use a spreadsheet with Expenses and Review_Queue tabs (columns matching the workflow fields) and connect Google Sheets OAuth2 credentials.