Telegram Bot Meal Calorie Estimator with Gemini & n8n AI Agent

Estimate meal calories in Telegram—using Google Gemini image understanding + an n8n AI Agent

This n8n workflow turns your Telegram bot into a meal calorie estimator. Authorized users can send meal details as text or photos, and the bot responds with an AI-calculated calorie estimate right inside Telegram—powered by Google Gemini and an n8n AI Agent.

What this workflow does

• Telegram trigger: The workflow starts when your Telegram bot receives a message.
• Access control: It checks the sender’s Telegram chat ID against an allowed chat ID. If it doesn’t match, the bot replies with an unauthorized message.
• Message routing: It routes content by type—handling text and photos, while rejecting unsupported types (for example, voice messages).
• Photo analysis with Gemini: For photos, the workflow downloads the image from Telegram and uses Google Gemini to generate a detailed meal description from the image.
• AI calorie estimation with an n8n AI Agent: It combines the user’s text (if provided) with the Gemini image description, then sends both to an n8n AI Agent that can use a calculator tool and conversation memory.
• Telegram response: The workflow sends the AI Agent’s calorie estimate back to the user in Telegram.

Use cases

• SaaS operators building a lightweight nutrition assistant inside Telegram for their users.
• Automation engineers prototyping “image-to-nutrition” experiences using n8n and Gemini.
• Personal diet tracking via a bot that accepts meal photos and returns calorie estimates.
• Authorized internal tools where only specific Telegram chat IDs can access the estimator.

Technical details

• Integrations: Telegram (trigger + send/download actions) and Google Gemini (image analysis + Gemini chat model for the AI Agent).
• Workflow logic: if, set, switch, aggregate, and routing with sticky note.
• Setup steps: Configure Telegram API credentials in n8n, add Google Gemini credentials, and set the authorized chat ID in the “Edit Fields” step. Optional: add an audio transcription step if you want voice support; otherwise keep the unsupported-message reply.