Invoice AI Agent – Telegram-Based Invoice Processing Automation
📌 Project Overview
A solo founder was handling invoices by hand through chat and email, with no central record of any of them.
I built the n8n agent that reads any invoice sent to Telegram, extracts the data, files it and replies instantly.
🚩 The Challenge
• Manual invoice checking was time-consuming
• High risk of human error when extracting totals and due dates
• No centralized invoice storage
• Slow response time to submitted invoices
📌 The Results
✔ Reduced invoice processing time from minutes to seconds
✔ Automated data extraction with consistent formatting
✔ Centralized invoice storage in Google Drive & Sheets
✔ Instant Telegram responses with summarized invoice details
✔ Improved accuracy and reduced manual workload
💡 The Solution
I built an end-to-end Invoice AI Agent using n8n that triggers when a user submits an invoice via Telegram. The workflow automatically downloads the file, analyzes the invoice image using AI vision and text parsing, extracts key details (total, due date, notes), and stores both the file and structured data in Google Drive and Google Sheets.
As the automation owner, I designed the workflow logic, AI prompt structure, and integrations to ensure fast, reliable, and user-friendly invoice handling.
🛠 Technologies Used
n8n, Telegram Bot API, Google Gemini Vision Model, OCR & Image Analysis, Google Drive API, Google Sheets, AI Agent with Memory Buffer
