Bibliothek / Finanzen, Rechnungen & Buchhaltung
Finanzen, Rechnungen & Buchhaltung
Telegram-Belegfotos mit Gemini-Ausgabenerkennung in Google Sheets protokollieren
Läuft, wenn ein Beleg an einen Telegram-Bot gesendet wird, nutzt Gemini, um Händler, Betrag, Datum und Kategorie zu extrahieren, speichert sichere Einträge in Google Sheets und bittet Nutzer, Einträge mit geringer Sicherheit zu bestätigen.
Was sie bewirkt
## AI Receipt Processor ### How it works This workflow lets users send receipt photos or documents to a Telegram bot. Gemini extracts the vendor, amount, date, expense category, payment method, and confidence score. High-confidence expenses are saved directly to Google Sheets. Low-confidence expenses are stored temporarily in an n8n Data Table and the user is asked to confirm or change the category in Telegram. ### Requirements - Telegram bot credentials - Google Gemini credentials - Google Sheets credentials - an n8n Data Table named pending_expenses - a Google Sheet for storing expenses ### Setup steps 1. Connect Telegram credentials to the trigger and all Telegram nodes. 2. Connect Google Gemini credentials to both Gemini nodes and select a Gemini model. 3. Make a copy of the provided Expense template and select it in all three Google Sheets nodes - https://docs.google.com/spreadsheets/d/129ZEubOfaVA1pOjbVCb9GgZQXbC48XTGpH7Zve5hBU0/edit?usp=sharing 4. Map these Google Sheets columns: - Date - Vendor - Amount - Payment type - Expense Category 5. Create an n8n Data Table named pending_expenses with: - request_id — Number - resume_url — String - vendor — String - amount — Number - date — Date/Time - category — String - payment_method — String 6. Select this Data Table in all Data Table nodes. 7. If you are self-hosting n8n, ensure your instance is publicly reachable for Telegram updates and wait/resume URLs. 8. Activate the workflow and send a receipt photo or document to your Telegram bot. ### Customization You can customize the confidence threshold, Gemini extraction prompt, expense categories, Telegram messages, Google Sheets columns, and category confirmation buttons. The default confidence threshold is 0.65. The default categories are Food, Transport, Shopping, Services, and Other.
Die Schritte
- Telegram Message Trigger
- Route by Telegram Input
- Download Telegram Photo
- Download Telegram Document
- Handle Unsupported Text Input
- Analyze Receipt Photo with Gemini
- Analyze Receipt Document with Gemini
- Parse Photo Analysis JSON
- Parse Document Analysis JSON
- Check Photo Analysis Confidence
- Check Document Analysis Confidence
- Append Photo Expense to Sheets
- Append Document Expense to Sheets
- Save Pending Expense to Data Table
Insgesamt 28 Schritte.
Ähnliche Automatisierungen
Datensynchronisierung & ETLSnowflake-Tabelle erstellen, Daten einfügen und aktualisieren6 SchritteE-Mail-AutomatisierungArchive IMAP email attachments to Folderit with verified data table receipts37 SchritteCode, Git & EntwicklertoolsCompare Shopify product variant changes with RapidAPI and a data table14 SchritteMonitoring, Betrieb & Störungenn8n-Workflow-Fehlermeldungen an Discord senden und in Data Table protokollieren7 Schritte