Skip to main content

Generate recipes from fridge photos using GPT-4 Vision & Telegram

Workflow preview

Workflow preview
100%
Generate recipes from fridge photos using GPT-4 Vision & Telegram preview
Open on n8n.io

Important notice

This workflow is provided as-is. Please review and test before using in production.

1. Workflow Overview

This n8n template demonstrates a multi modal AI recipe assistant that suggests delicious recipes based on user input, delivered via Telegram. The workflow can uniquely handle two types of input: a ...

Best for

  • Content Creation automation workflows
  • Multimodal AI automation workflows
  • advanced n8n builders looking for reusable templates

Tools used

n8n-nodes-base.if, @n8n/n8n-nodes-langchain.agent, @n8n/n8n-nodes-langchain.lmchatopenrouter, @n8n/n8n-nodes-langchain.outputparserstructured, n8n-nodes-base.set, n8n-nodes-base.telegramtrigger, n8n-nodes-base.telegram, n8n-nodes-base.stickynote

Source and attribution

This workflow is cataloged by N8N Workflows and links back to its original n8n.io source page by Yusuke Yamamoto.

Original n8n.io source

1.1 Workflow description

Title
Generate recipes from fridge photos using GPT-4 Vision & Telegram
Workflow name
Generate recipes from fridge photos using GPT-4 Vision & Telegram

This n8n template demonstrates a multi-modal AI recipe assistant that suggests delicious recipes based on user input, delivered via Telegram. The workflow can uniquely handle two types of input: a photo of your ingredients or a simple text list.

Use cases are many: Get instant dinner ideas by taking a photo of your fridge contents, reduce food waste by finding recipes for leftover ingredients, or create a fun and interactive service for a cooking community or food delivery app!

Good to know

  • This workflow uses two different AI models (one for vision, one for text generation), so costs will be incurred for each execution. See OpenRouter Pricing or your chosen model provider's pricing page for updated info.
  • The AI prompts are in English, but the final recipe output is configured to be in Japanese. You can easily change the language by editing the prompt in the Recipe Generator node.

How it works

  1. The workflow starts when a user sends a message or an image to your bot on Telegram via the Telegram Trigger.
  2. An IF node intelligently checks if the input is text or an image.
  3. If an image is sent, the AI Vision Agent analyzes it to identify ingredients. A Structured Output Parser then forces this data into a clean JSON list.
  4. If text is sent, a Set node directly prepares the user's text as the ingredient list.
  5. Both paths converge, providing a standardized ingredient list to the Recipe Generator agent. This AI acts as a professional chef to create three detailed recipes.
  6. Crucially, a second Structured Output Parser takes the AI's creative text and formats it into a reliable JSON structure (with name, difficulty, instructions, etc.). This ensures the output is always predictable and easy to work with.
  7. A final Set node uses a JavaScript expression to transform the structured recipe data into a beautiful, emoji-rich, and easy-to-read message.
  8. The formatted recipe suggestions are sent back to the user on Telegram.

How to use

  • Configure the Telegram Trigger with your own bot's API credentials.
  • Add your AI provider credentials in the OpenAI Vision Model and OpenAI Recipe Model nodes (this template uses OpenRouter, but it can be swapped for a direct OpenAI connection).

Requirements

  • A Telegram account and a bot token.
  • An AI provider account that supports vision and text models, such as OpenRouter or OpenAI.

Customising this workflow

  • Modify the prompt in the Recipe Generator to include dietary restrictions (e.g., "vegan," "gluten-free") or to change the number of recipes suggested.
  • Swap the Telegram nodes for Discord, Slack, or a Webhook to integrate this recipe bot into a different platform or your own application.
  • Connect to a recipe database API to supplement the AI's suggestions with existing recipes.

1.2 Logical Blocks

This catalog entry is organized from the workflow JSON. The node-level section below shows the executable blocks available for review before importing the template.

2. Block-by-Block Analysis

Block 1 - Check Input Type

Type / Role
n8n-nodes-base.if - if
Config choices
Version 2

Block 2 - AI Vision Agent

Type / Role
@n8n/n8n-nodes-langchain.agent - agent
Config choices
Version 2

Block 3 - OpenAI Vision Model

Type / Role
@n8n/n8n-nodes-langchain.lmChatOpenRouter - lmChatOpenRouter
Config choices
Version 1

Block 4 - Ingredient Parser

Type / Role
@n8n/n8n-nodes-langchain.outputParserStructured - outputParserStructured
Config choices
Version 1.2

Block 5 - Format Ingredients

Type / Role
n8n-nodes-base.set - set
Config choices
Version 3.4

Block 6 - Get Text Ingredients

Type / Role
n8n-nodes-base.set - set
Config choices
Version 3.4

Block 7 - Recipe Generator

Type / Role
@n8n/n8n-nodes-langchain.agent - agent
Config choices
Version 2

Block 8 - OpenAI Recipe Model

Type / Role
@n8n/n8n-nodes-langchain.lmChatOpenRouter - lmChatOpenRouter
Config choices
Version 1

Block 9 - Recipe Parser

Type / Role
@n8n/n8n-nodes-langchain.outputParserStructured - outputParserStructured
Config choices
Version 1.2

Block 10 - Format Response

Type / Role
n8n-nodes-base.set - set
Config choices
Version 3.4

Block 11 - Telegram Trigger

Type / Role
n8n-nodes-base.telegramTrigger - telegramTrigger
Config choices
Version 1.2

Block 12 - Send a text message

Type / Role
n8n-nodes-base.telegram - telegram
Config choices
Version 1.2

Block 13 - Sticky Note

Type / Role
n8n-nodes-base.stickyNote - stickyNote
Config choices
Version 1

Block 14 - Sticky Note1

Type / Role
n8n-nodes-base.stickyNote - stickyNote
Config choices
Version 1

Block 15 - Sticky Note2

Type / Role
n8n-nodes-base.stickyNote - stickyNote
Config choices
Version 1

Block 16 - Sticky Note3

Type / Role
n8n-nodes-base.stickyNote - stickyNote
Config choices
Version 1

3. Summary Table

Workflow Generate recipes from fridge photos using GPT-4 Vision & Telegram
Complexity advanced
Nodes 16
Categories Content Creation, Multimodal AI
Author Yusuke Yamamoto
Published 13 Oct 2025

4. Reproducing the Workflow from Scratch

  1. 1. Download the workflow JSON

    Use the JSON export at /data/workflows/9558/9558.json as the source template for this automation.

  2. 2. Import the template into n8n

    Open n8n, import the downloaded JSON, and review each node before activating the workflow.

  3. 3. Configure credentials and variables

    Replace placeholder credentials, API keys, webhook URLs, account IDs, and environment-specific values with your own settings.

  4. 4. Test with sample data

    Run the workflow manually or in a staging workspace, inspect node output, and confirm downstream systems receive the expected data.

  5. 5. Activate and monitor

    Enable the workflow only after testing, then monitor executions, errors, and rate limits during the first production runs.

5. General Notes & Resources

Review imported nodes carefully before activation. This catalog entry is intended to help you inspect the workflow structure, understand required services, and find related templates faster.

Node names, credentials, schedules, webhook paths, and external service limits may need adjustment for your workspace.

Frequently asked questions

What does Generate recipes from fridge photos using GPT-4 Vision & Telegram do?

This n8n template demonstrates a multi modal AI recipe assistant that suggests delicious recipes based on user input, delivered via Telegram. The workflow can uniquely handle two types of input: a ...

What do I need before importing this workflow?

Review the workflow JSON, configure any required credentials in n8n, and test the automation in a safe workspace before using it in production.

Can I customize this workflow?

Yes. Use the block-by-block analysis and the downloadable JSON to inspect each node, then adjust credentials, prompts, schedules, filters, or destinations for your Content Creation, Multimodal AI use case.