January 22, 2026

Telegram + OpenAI: researched posts from voice notes

Lisa Granqvist Partner Workflow Automation Expert

Get a free AI assessment → ⬇️ Use template

Creating “one quick post” shouldn’t turn into 12 tabs, three half-finished drafts, and a note to yourself to “add sources later.” But that’s what happens when research, ideation, and writing all live in different places.

Social media managers feel this every day. A marketing lead trying to keep a weekly cadence feels it too. And honestly, so does the founder who just wants Telegram OpenAI posts without learning a new tool. This workflow turns a voice note into researched drafts and image prompts in one pass.

Below you’ll see exactly how the automation runs, what you get out the other end, and where teams usually customize it to match their brand voice.

How This Automation Works

The full n8n workflow, from trigger to final output:

n8n Workflow Template: Telegram + OpenAI: researched posts from voice notes

Click to explore

flowchart LR

    subgraph sg0["Receive Telegram Messages Flow"]
        direction LR
        n0["<div style='background:#f5f5f5;padding:10px;border-radius:8px;display:inline-block;border:1px solid #e0e0e0'><img src='https://flowpast.com/wp-content/uploads/n8n-workflow-icons/telegram.svg' width='40' height='40' /></div><br/>Receive Telegram Messages"]
        n1@{ icon: "mdi:swap-horizontal", form: "rounded", label: "Voice or Text?", pos: "b", h: 48 }
        n2["<div style='background:#f5f5f5;padding:10px;border-radius:8px;display:inline-block;border:1px solid #e0e0e0'><img src='https://flowpast.com/wp-content/uploads/n8n-workflow-icons/telegram.svg' width='40' height='40' /></div><br/>Fetch Voice Message"]
        n3@{ icon: "mdi:robot", form: "rounded", label: "Transcribe Voice to Text", pos: "b", h: 48 }
        n4@{ icon: "mdi:swap-vertical", form: "rounded", label: "Prepare for LLM", pos: "b", h: 48 }
        n5@{ icon: "mdi:robot", form: "rounded", label: "AI Agent", pos: "b", h: 48 }
        n6@{ icon: "mdi:brain", form: "rounded", label: "OpenAI Chat Model", pos: "b", h: 48 }
        n7@{ icon: "mdi:wrench", form: "rounded", label: "SerpAPI", pos: "b", h: 48 }
        n8@{ icon: "mdi:robot", form: "rounded", label: "Structured Output Parser", pos: "b", h: 48 }
        n9@{ icon: "mdi:cog", form: "rounded", label: "Extract from File", pos: "b", h: 48 }
        n10@{ icon: "mdi:swap-vertical", form: "rounded", label: "Prepare Final Output", pos: "b", h: 48 }
        n11["<div style='background:#f5f5f5;padding:10px;border-radius:8px;display:inline-block;border:1px solid #e0e0e0'><img src='https://flowpast.com/wp-content/uploads/n8n-workflow-icons/httprequest.dark.svg' width='40' height='40' /></div><br/>Generate Image"]
        n7 -.-> n5
        n5 --> n11
        n11 --> n9
        n1 --> n2
        n1 --> n4
        n4 --> n5
        n9 --> n10
        n6 -.-> n5
        n2 --> n3
        n8 -.-> n5
        n3 --> n5
        n0 --> n1
    end

    %% Styling
    classDef trigger fill:#e8f5e9,stroke:#388e3c,stroke-width:2px
    classDef ai fill:#e3f2fd,stroke:#1976d2,stroke-width:2px
    classDef aiModel fill:#e8eaf6,stroke:#3f51b5,stroke-width:2px
    classDef decision fill:#fff8e1,stroke:#f9a825,stroke-width:2px
    classDef database fill:#fce4ec,stroke:#c2185b,stroke-width:2px
    classDef api fill:#fff3e0,stroke:#e65100,stroke-width:2px
    classDef code fill:#f3e5f5,stroke:#7b1fa2,stroke-width:2px
    classDef disabled stroke-dasharray: 5 5,opacity: 0.5
    class n0 trigger
    class n3,n5,n8 ai
    class n6 aiModel
    class n7 ai
    class n1 decision
    class n11 api
    classDef customIcon fill:none,stroke:none
    class n0,n2,n11 customIcon

The Problem: social post drafting still takes too long

You already know the pattern. You have a solid topic idea, but then you need “just a little research,” which becomes a rabbit hole. Next, you open a doc, start drafting, second-guess the hook, and realize you still need an image concept. Then it’s back to Google, back to your notes app, back to your prompt library. By the time you’ve got something decent, you’ve spent about an hour and your energy is gone. The worst part is the inconsistency: some posts are great, others feel rushed because the process is hard to repeat.

The friction compounds. Here’s where it usually breaks down.

Research and writing happen in separate tools, so you keep copying snippets back and forth.
Voice memos are fast to record, but turning them into usable drafts still becomes a manual chore.
Image direction is an afterthought, which means visuals look generic or you skip them entirely.
When you’re busy, quality drops because the process depends on your focus and time that day.

The Solution: turn Telegram messages into researched drafts

This workflow lets you message a Telegram bot with either text or a voice note, then it does the heavy lifting for you. If you send audio, it grabs the voice clip and transcribes it with OpenAI Whisper. Next, the workflow packages your request into a clean prompt and hands it to an AI agent that can do real online research using SerpAPI. Based on what it finds, the agent produces structured, ready-to-edit social post drafts plus a detailed image prompt you can use in your preferred art tool. If you enable the optional image step, it can even send a request out via HTTP to generate or fetch an image and attach it to the final output.

It starts with a simple Telegram message. AI handles transcription (if needed), research, and drafting. Finally, you receive a tidy result bundle you can paste into your scheduler, send for approval, or store for later.

What You Get: Automation vs. Results

What This Workflow Automates

Results You’ll Get

It listens for Telegram text or voice messages and routes them automatically.
Voice notes get transcribed into usable text with OpenAI Whisper.
SerpAPI-powered research is pulled in before drafts are written.
It generates both social post drafts and detailed image prompts in one run.

Most teams get a first draft in about 5 minutes, not an hour.
Your posts reference real context from online research instead of “vibes only.”
You ship more consistently because the workflow doesn’t get tired.
Image prompts are specific enough to hand to a designer or an AI image tool.
You reduce last-minute scrambling because you can create drafts in batches.

Example: What This Looks Like

Say you publish 5 posts a week. Manually, a “quick” research + draft pass is often about 1 hour per post, once you include reading sources, writing, and coming up with an image concept (so roughly 5 hours weekly). With this workflow, you send a 30-second voice note in Telegram, wait a few minutes for transcription, research, and drafting, and you’re done. Call it about 5 minutes of your time per post. That’s roughly 4 hours back every week, without lowering the bar.

What You’ll Need

n8n instance (try n8n Cloud free)
Self-hosting option if you prefer (Hostinger works well)
Telegram for capturing voice notes and text prompts.
OpenAI to transcribe audio and generate drafts.
SerpAPI API key (get it from your SerpAPI dashboard).

Skill level: Intermediate. You’ll connect credentials, paste API keys, and test a few runs in Telegram.

Don’t want to set this up yourself? Talk to an automation expert (free 15-minute consultation).

How It Works

Telegram message kicks it off. You send a text prompt or a voice note to your Telegram bot, and n8n instantly triggers the workflow.

Voice messages become text. If the incoming message is audio, the workflow fetches the voice clip from Telegram and sends it to OpenAI Whisper for transcription, so the rest of the automation can treat voice and text the same way.

Research + drafting happen together. Your request is packaged into a clean input, then the AI agent uses an OpenAI chat model plus a SerpAPI research tool to pull context and produce structured outputs (draft posts and an image prompt).

Final output is assembled for reuse. n8n formats everything into a consistent response. If you enable the image step, an HTTP request can generate or retrieve an image and pass the binary file into the final result.

You can easily modify the research sources and output format to match each platform. See the full implementation guide below for customization options.

Step-by-Step Implementation Guide

Step 1: Configure the Telegram Trigger

Start by setting up the incoming Telegram message trigger so the workflow can receive voice or text messages.

Add and configure Telegram Intake Trigger as your entry node.
Set Updates to include message.
Credential Required: Connect your telegramApi credentials in Telegram Intake Trigger.
Leave Additional Fields empty unless you need advanced Telegram options.

The Flowpast Branding sticky note is informational only and does not affect execution.

Step 2: Route Voice vs Text Inputs

Use a switch to separate audio messages from text messages before processing.

Configure Route Voice vs Text with three outputs: Audio, Text, and Error.
Set the Audio rule to check existence of ={{ $json.message.voice.file_id }}.
Set the Text rule to check existence of ={{ $json.message.text || "" }}.
Leave the Error output as a placeholder for future error handling.

⚠️ Common Pitfall: If your Telegram bot sends other update types, the switch may route to Error. Keep updates limited to message in Telegram Intake Trigger.

Step 3: Process Voice and Text Inputs

Convert audio messages to text and normalize plain text input for the AI agent.

On the Audio path, configure Retrieve Voice Clip with Resource set to file and File ID set to ={{ $json.message.voice.file_id }}.
Credential Required: Connect your telegramApi credentials in Retrieve Voice Clip.
Connect Retrieve Voice Clip to Convert Audio to Text and set Resource to audio and Operation to translate.
Credential Required: Connect your openAiApi credentials in Convert Audio to Text.
On the Text path, configure Assemble LLM Input to assign text to ={{$json.message.text}}.

Step 4: Set Up the AI Agent and Tools

Configure the AI agent, its model, research tool, and structured output parser.

Configure Content Strategy Agent with Text set to ={{$json.text}} and Prompt Type set to define.
Ensure Content Strategy Agent has Has Output Parser enabled.
Connect OpenAI Chat Engine as the language model for Content Strategy Agent and set Model to gpt-4o-mini.
Credential Required: Connect your openAiApi credentials in OpenAI Chat Engine.
Connect Serp Research Tool as the tool for Content Strategy Agent.
Credential Required: Connect your serpApi credentials in Serp Research Tool (this is a tool sub-node for Content Strategy Agent).
Connect Structured JSON Parser as the output parser for Content Strategy Agent and keep the schema example:

{ "content": "[SOCIAL_MEDIA_CONTENT]", "image_prompt": "[IMAGE_PROMPT]" }

The AI tool and parser nodes rely on Content Strategy Agent. Add or update credentials on the parent/linked nodes (OpenAI Chat Engine, Serp Research Tool), not on the parser itself.

Step 5: Generate Image and Compose the Final Output

Send the image prompt to Hugging Face, extract the binary response, and assemble the final payload.

Configure Create Image Request with URL set to https://router.huggingface.co/hf-inference/models/stabilityai/stable-diffusion-3.5-large and Method set to POST.
Enable Send Body and set inputs to ={{ $json.output.image_prompt }}.
Credential Required: Connect your huggingFaceApi credentials in Create Image Request.
If required by your endpoint, also connect httpHeaderAuth credentials in Create Image Request.
Configure Extract Binary Content with Operation set to binaryToPropery to transform the image output.
Configure Compose Final Result to assign content to ={{ $('Content Strategy Agent').item.json.output.content }} and image to ={{ $json.data }}.

⚠️ Common Pitfall: If the Hugging Face model returns an error, Extract Binary Content may not produce data. Validate the model endpoint and authentication before production use.

Step 6: Test and Activate Your Workflow

Verify the end-to-end flow from Telegram input to final output and then activate the workflow.

Click Execute Workflow and send a test message (text or voice) to your Telegram bot.
Confirm Route Voice vs Text routes correctly and Convert Audio to Text creates transcribed text for voice inputs.
Verify Content Strategy Agent returns JSON with content and image_prompt, and that Create Image Request returns image data.
Check Compose Final Result for a combined payload containing the generated text and image data.
Once validated, toggle the workflow to Active for production use.

🔒

Unlock Full Step-by-Step Guide

Get the complete implementation guide + downloadable template

Common Gotchas

Telegram bot tokens can be revoked or you may be messaging the wrong bot. If intake suddenly stops, check the Telegram Trigger node settings and the bot token first.
If you’re using Wait nodes or external rendering, processing times vary. Bump up the wait duration if downstream nodes fail on empty responses.
Default prompts in AI nodes are generic. Add your brand voice early or you’ll be editing outputs forever.

Frequently Asked Questions

How long does it take to set up this Telegram OpenAI posts automation?

About 30 minutes if your Telegram bot and API keys are ready.

Do I need coding skills to automate Telegram OpenAI posts?

No. You’ll mainly connect accounts and paste API keys into n8n. The “logic” is already built into the workflow.

Is n8n free to use for this Telegram OpenAI posts workflow?

Yes. n8n has a free self-hosted option and a free trial on n8n Cloud. Cloud plans start at $20/month for higher volume. You’ll also need to factor in OpenAI and SerpAPI usage; for light drafting, many teams spend a few dollars a month.

Where can I host n8n to run this automation?

Two options: n8n Cloud (managed, easiest setup) or self-hosting on a VPS. For self-hosting, Hostinger VPS is affordable and handles n8n well. Self-hosting gives you unlimited executions but requires basic server management.

Can I customize this Telegram OpenAI posts workflow for LinkedIn-style posts instead of Instagram captions?

Yes, and it’s worth doing. Update the instructions you send into the Content Strategy Agent so it writes in a LinkedIn voice (tighter hook, fewer hashtags, clearer takeaway). Many teams also adjust the Structured JSON Parser output so it returns multiple variations: one short, one story-driven, and one “contrarian” angle. If you keep the structure consistent, approvals get easier because everyone knows what they’re reviewing.

Why is my Telegram connection failing in this workflow?

Usually it’s a bad bot token or the bot was removed from the chat you’re testing in. Regenerate the token in BotFather, then update the Telegram credentials in n8n. If voice notes fail but text works, check the “Retrieve Voice Clip” node permissions and confirm the workflow is receiving audio files from your Telegram client.

How many requests can this Telegram OpenAI posts automation handle?

On a small n8n Cloud plan, it comfortably handles dozens of requests a day for most teams.

Is this Telegram OpenAI posts automation better than using Zapier or Make?

It depends on how “smart” you need the middle part to be. This workflow uses an AI agent with research tooling and structured outputs, which is much easier to control in n8n when you want branching and richer logic. n8n also gives you a self-hosted path, so you’re not paying per tiny step once volume ramps up. Zapier or Make can still be fine for simple “message in, message out” flows, and they may feel simpler on day one. If you’re unsure, Talk to an automation expert and we’ll map the cheapest option for your usage.

Set it up once, then create drafts the moment an idea hits. The workflow handles the repetitive parts so you can focus on publishing and improving what works.

Need Help Setting This Up?

Our automation experts can build and customize this workflow for your specific needs. Free 15-minute consultation—no commitment required.

Lisa Granqvist

Workflow Automation Expert

Expert in workflow automation and no-code tools.

{
"id": null,
"name": "Automated Telegram Content Creator",
"meta": {
"instanceId": "0084a61d8c3b517c7fffb4cca2959f5ddbc6576e26b330d1d41c4e9de4365f28",
"templateCredsSetupCompleted": null,
"templateId": null
},
"nodes": [
{
"id": "flowpast-topbar-3057",
"name": "Flowpast Branding",
"type": "n8n-nodes-base.stickyNote",
"position": [
0,
20
],
"parameters": {
"color": 7,
"width": 1190,
"height": 80,
"content": "## Flowpast.com | Automation Workflow Library\n**\ud83d\udcd6 Full tutorial & setup guide:** flowpast.com"
},
"typeVersion": 1
},
{
"id": "070fd7b4-58ca-4372-a347-6f60f590e20b",
"name": "Telegram Intake Trigger",
"type": "n8n-nodes-base.telegramTrigger",
"position": [
1090,
150
],
"webhookId": "4e2cd560-ae4e-4ed7-a8ea-984518404e51",
"parameters": {
"updates": [
"message"
],
"additionalFields": []
},
"credentials": {
"telegramApi": {
"id": "credential-id",
"name": ""
}
},
"typeVersion": 1.1
},
{
"id": "0e0f3a32-fbde-42a9-aa7f-70fda7b05357",
"name": "Route Voice vs Text",
"type": "n8n-nodes-base.switch",
"position": [
795,
240
],
"parameters": {
"rules": {
"values": [
{
"outputKey": "Audio",
"conditions": {
"options": {
"version": 2,
"leftValue": "",
"caseSensitive": true,
"typeValidation": "strict"
},
"combinator": "and",
"conditions": [
{
"id": "af30c479-4542-405f-b315-37c50c4e2bef",
"operator": {
"type": "string",
"operation": "exists",
"singleValue": true
},
"leftValue": "={{ $json.message.voice.file_id }}",
"rightValue": ""
}
]
},
"renameOutput": true
},
{
"outputKey": "Text",
"conditions": {
"options": {
"version": 2,
"leftValue": "",
"caseSensitive": true,
"typeValidation": "strict"
},
"combinator": "and",
"conditions": [
{
"id": "a3ca8cd4-fbb2-40b5-829a-24724f2fbc85",
"operator": {
"type": "string",
"operation": "exists",
"singleValue": true
},
"leftValue": "={{ $json.message.text || \"\" }}",
"rightValue": ""
}
]
},
"renameOutput": true
},
{
"outputKey": "Error",
"conditions": {
"options": {
"version": 2,
"leftValue": "",
"caseSensitive": true,
"typeValidation": "strict"
},
"combinator": "and",
"conditions": [
{
"id": "9bcfdee0-2f09-4037-a7b9-689ef392371d",
"operator": {
"type": "string",
"operation": "exists",
"singleValue": true
},
"leftValue": "error",
"rightValue": ""
}
]
},
"renameOutput": true
}
]
},
"options": []
},
"typeVersion": 3.2
},
{
"id": "b01dde88-bede-4500-974f-b2dc203ff841",
"name": "Retrieve Voice Clip",
"type": "n8n-nodes-base.telegram",
"position": [
640,
120
],
"webhookId": "23645237-4943-4c32-b18c-97c410cc3409",
"parameters": {
"fileId": "={{ $json.message.voice.file_id }}",
"resource": "file"
},
"credentials": {
"telegramApi": {
"id": "credential-id",
"name": ""
}
},
"typeVersion": 1.2
},
{
"id": "fe91414e-3b10-482e-b8dd-d55266828dd7",
"name": "Convert Audio to Text",
"type": "@n8n/n8n-nodes-langchain.openAi",
"position": [
350,
155
],
"parameters": {
"options": [],
"resource": "audio",
"operation": "translate"
},
"credentials": {
"openAiApi": {
"id": "credential-id",
"name": ""
}
},
"typeVersion": 1.8
},
{
"id": "74549458-fd4d-4824-a561-944f2f536b9b",
"name": "Assemble LLM Input",
"type": "n8n-nodes-base.set",
"position": [
620,
445
],
"parameters": {
"options": [],
"assignments": {
"assignments": [
{
"id": "b324a329-3c49-4f7f-b683-74331b7fe7f8",
"name": "=text",
"type": "string",
"value": "={{$json.message.text}}"
}
]
}
},
"typeVersion": 3.4
},
{
"id": "156580f1-adf5-43ba-b54d-89b84ca87818",
"name": "Content Strategy Agent",
"type": "@n8n/n8n-nodes-langchain.agent",
"position": [
180,
260
],
"parameters": {
"text": "={{$json.text}}",
"options": {
"systemMessage": " \n**1. AI Agent Goal Prompt (Overall Task)**\n\n*   **Purpose:** To define the agent's overall objective, removing the image creation step.\n\n```\nYou are an AI social media content creator. Your task is to research a given topic using SerpAPI, create engaging and SEO-optimized social media content (800-1000 characters), and generate a detailed image prompt. The content must be factually accurate and engaging. Prioritize factual accuracy and engaging storytelling in your content. The generated image prompt should be detailed and specific enough to be used with an image generation tool like DALL-E or Stable Diffusion.\n```\n\n**2. SerpAPI Tool Prompt (Research Phase)**\n\n*   **Purpose:** To instruct the agent on how to use SerpAPI to effectively gather information. (No change from previous version)\n\n```\nUse the SerpAPI tool to research the following topic: [TOPIC]. Focus on identifying key facts, trends, and interesting angles relevant for social media. Extract information from the top search results. Return a summary of the information you found focusing on key data and facts.\n```\n\n*   **Explanation:**\n    *   `[TOPIC]` is a variable that will be replaced with the specific topic.\n    *   Focuses the agent on extracting key facts and trends rather than just providing a list of results.\n    *   Limits the scope to the top search results to maintain efficiency.\n\n**3. Social Media Content Creation Prompt**\n\n*   **Purpose:** To guide the agent in creating engaging and SEO-friendly content based on the research. (No change from previous version)\n\n```\nBased on the following research summary: [RESEARCH_SUMMARY], create a social media post that is:\n\n*   Engaging and attention-grabbing\n*   Factually accurate\n*   Optimized for SEO (include relevant keywords naturally)\n*   Within 800-1000 characters\n*   Clearly and concisely written.\n*   Avoid jargon and technical terms.\n*    Include a call to action.\n\nThe tone should be informative but also enthusiastic and easily understandable.\n```\n\n*   **Explanation:**\n    *   `[RESEARCH_SUMMARY]` will be replaced with the output from the SerpAPI tool.\n    *   Specific instructions on tone, length, and SEO optimization.\n    *   Explicitly asks for clear and concise writing, avoiding jargon.\n    *   Added \"Include a call to action\" to make the content more actionable\n\n**4. Image Generation Prompt (for hypothetical image generation tool)**\n\n*   **Purpose:** To create a prompt that generates a detailed and descriptive image prompt for an image generation tool.  This prompt should now be the *final output* related to the image.\n\n```\nBased on the following topic: [TOPIC] and social media content: [SOCIAL_MEDIA_CONTENT], generate a detailed image prompt for a photorealistic image that visually represents the topic and complements the content. The image should be:\n\n*   Photorealistic and high-quality.\n*   Visually appealing and attention-grabbing.\n*   Relevant to the topic and content.\n*   Appropriate for social media.\n\nThe prompt should be exceptionally detailed and specific, providing precise instructions for an image generation tool like DALL-E or Stable Diffusion. Include details about the subject, setting, style, lighting, camera angles, and any other relevant visual elements.  Aim for a prompt that leaves no room for misinterpretation by the image generation AI.  Mention specific artists or photographic styles to emulate if appropriate.\n```\n\n*   **Explanation:**\n    *   `[TOPIC]` and `[SOCIAL_MEDIA_CONTENT]` are variables that will be replaced with the topic and the created social media content, respectively.\n    *   Focuses on *photorealism*, relevance, and visual appeal.\n    *   Emphasizes the need for an *exceptionally detailed and specific* prompt for the image generation tool.\n    *  Explicitly mentions DALL-E and Stable Diffusion as target tools.\n    *   Advises the inclusion of artist styles or photographic techniques to guide the image generation.\n\n**5. JSON Output Instruction**\n\n*   **Purpose:** To ensure the AI agent provides the output in the correct format.  The `image_url` field is replaced with `image_prompt`.\n\n```\nAfter generating the social media content and the image prompt, output the results in the following JSON format:\n\n```json\n{\n\"content\": \"[SOCIAL_MEDIA_CONTENT]\",\n\"image_prompt\": \"[IMAGE_PROMPT]\"\n}\n```\n\n`[SOCIAL_MEDIA_CONTENT]` is the social media content you created.\n`[IMAGE_PROMPT]` is the detailed image prompt you generated.\n```\n\n**Example Usage:**\n\nLet's say the topic is still \"The Benefits of Regular Exercise.\"\n\n1.  **SerpAPI Tool:** The agent uses SerpAPI to find information about the benefits of exercise.\n2.  **Social Media Content:** The agent generates content like: \"Boost your mood & health! \ud83d\udcaa Regular exercise reduces stress, improves sleep, and lowers disease risk. Get moving today! #exercise #healthylifestyle #fitness\"\n3.  **Image Prompt:** The agent generates an image prompt like: \"A photorealistic image of a diverse group of people happily participating in various forms of exercise in a vibrant outdoor setting. Some are jogging in a park with lush green trees, others are doing yoga poses on a grassy field, and a few are cycling on a paved path. The lighting is warm and golden, as if it's early morning or late afternoon. The style should be reminiscent of a National Geographic photograph, emphasizing the natural beauty of the scene and the healthy glow of the people. Use a shallow depth of field to blur the background slightly, drawing focus to the subjects. Camera angle: slightly low, capturing the energy and movement of the scene. Consider influences from the photographic style of Steve McCurry.\"\n4.  **JSON Output:**\n\n```json\n{\n\"content\": \"Boost your mood & health! \ud83d\udcaa Regular exercise reduces stress, improves sleep, and lowers disease risk. Get moving today! #exercise #healthylifestyle #fitness\",\n\"image_prompt\": \"A photorealistic image of a diverse group of people happily participating in various forms of exercise in a vibrant outdoor setting. Some are jogging in a park with lush green trees, others are doing yoga poses on a grassy field, and a few are cycling on a paved path. The lighting is warm and golden, as if it's early morning or late afternoon. The style should be reminiscent of a National Geographic photograph, emphasizing the natural beauty of the scene and the healthy glow of the people. Use a shallow depth of field to blur the background slightly, drawing focus to the subjects. Camera angle: slightly low, capturing the energy and movement of the scene. Consider influences from the photographic style of Steve McCurry.\"\n}\n```\n\n**Key Improvements and Techniques Used (Beyond the Previous Version):**\n\n*   **Focus on Photorealism:**  The image prompt now explicitly aims for photorealistic results.\n*   **Detailed Image Prompting:** The prompt emphasizes extreme detail and specificity in the image prompt.\n*   **Tool Agnostic:** The prompt mentions DALL-E and Stable Diffusion as example tools, but is designed to be usable with other image generation AIs.\n*   **Artist Style Guidance:** The prompt encourages the inclusion of artist or photographic style references.\n\n \n"
},
"promptType": "define",
"hasOutputParser": true
},
"typeVersion": 1.7
},
{
"id": "74f99543-d7e6-4d9b-8af1-9d86e0566ddc",
"name": "OpenAI Chat Engine",
"type": "@n8n/n8n-nodes-langchain.lmChatOpenAi",
"position": [
185,
590
],
"parameters": {
"model": {
"__rl": true,
"mode": "list",
"value": "gpt-4o-mini"
},
"options": []
},
"credentials": {
"openAiApi": {
"id": "credential-id",
"name": ""
}
},
"typeVersion": 1.2
},
{
"id": "62b9bb5f-0c87-4df4-ac1c-70b96a0a5cc4",
"name": "Serp Research Tool",
"type": "@n8n/n8n-nodes-langchain.toolSerpApi",
"position": [
430,
530
],
"parameters": {
"options": []
},
"credentials": {
"serpApi": {
"id": "credential-id",
"name": ""
}
},
"typeVersion": 1
},
{
"id": "3782a100-3210-4c87-9e6a-5808cd488601",
"name": "Structured JSON Parser",
"type": "@n8n/n8n-nodes-langchain.outputParserStructured",
"position": [
0,
575
],
"parameters": {
"jsonSchemaExample": "{\n\"content\": \"[SOCIAL_MEDIA_CONTENT]\",\n\"image_prompt\": \"[IMAGE_PROMPT]\"\n}"
},
"typeVersion": 1.2
},
{
"id": "87bf4b00-4f75-4098-993b-b4bd168339c2",
"name": "Extract Binary Content",
"type": "n8n-nodes-base.extractFromFile",
"position": [
600,
310
],
"parameters": {
"options": [],
"operation": "binaryToPropery"
},
"typeVersion": 1
},
{
"id": "c907aa15-1ccf-475e-94da-3a81e54b3746",
"name": "Compose Final Result",
"type": "n8n-nodes-base.set",
"position": [
885,
265
],
"parameters": {
"options": [],
"assignments": {
"assignments": [
{
"id": "df5eb034-ef40-44a3-a620-48981efd1a69",
"name": "content",
"type": "string",
"value": "={{ $('Content Strategy Agent').item.json.output.content }}"
},
{
"id": "9ed8afc9-a957-4aea-8554-8c67017ef0e6",
"name": "image",
"type": "string",
"value": "={{ $json.data }}"
}
]
}
},
"typeVersion": 3.4
},
{
"id": "0893eac1-e72b-4a95-8c3c-4803aaaed9b9",
"name": "Create Image Request",
"type": "n8n-nodes-base.httpRequest",
"position": [
435,
300
],
"parameters": {
"url": "https://router.huggingface.co/hf-inference/models/stabilityai/stable-diffusion-3.5-large",
"method": "POST",
"options": [],
"sendBody": true,
"authentication": "predefinedCredentialType",
"bodyParameters": {
"parameters": [
{
"name": "inputs",
"value": "={{ $json.output.image_prompt }}"
}
]
},
"nodeCredentialType": "huggingFaceApi"
},
"credentials": {
"httpHeaderAuth": {
"id": "credential-id",
"name": ""
},
"huggingFaceApi": {
"id": "credential-id",
"name": ""
}
},
"typeVersion": 4.2
}
],
"pinData": [],
"connections": {
"Serp Research Tool": {
"ai_tool": [
[
{
"node": "Content Strategy Agent",
"type": "ai_tool",
"index": 0
}
]
]
},
"Content Strategy Agent": {
"main": [
[
{
"node": "Create Image Request",
"type": "main",
"index": 0
}
]
]
},
"Create Image Request": {
"main": [
[
{
"node": "Extract Binary Content",
"type": "main",
"index": 0
}
]
]
},
"Route Voice vs Text": {
"main": [
[
{
"node": "Retrieve Voice Clip",
"type": "main",
"index": 0
}
],
[
{
"node": "Assemble LLM Input",
"type": "main",
"index": 0
}
]
]
},
"Assemble LLM Input": {
"main": [
[
{
"node": "Content Strategy Agent",
"type": "main",
"index": 0
}
]
]
},
"Extract Binary Content": {
"main": [
[
{
"node": "Compose Final Result",
"type": "main",
"index": 0
}
]
]
},
"OpenAI Chat Engine": {
"ai_languageModel": [
[
{
"node": "Content Strategy Agent",
"type": "ai_languageModel",
"index": 0
}
]
]
},
"Retrieve Voice Clip": {
"main": [
[
{
"node": "Convert Audio to Text",
"type": "main",
"index": 0
}
]
]
},
"Compose Final Result": {
"main": [
[]
]
},
"Structured JSON Parser": {
"ai_outputParser": [
[
{
"node": "Content Strategy Agent",
"type": "ai_outputParser",
"index": 0
}
]
]
},
"Convert Audio to Text": {
"main": [
[
{
"node": "Content Strategy Agent",
"type": "main",
"index": 0
}
]
]
},
"Telegram Intake Trigger": {
"main": [
[
{
"node": "Route Voice vs Text",
"type": "main",
"index": 0
}
]
]
}
},
"versionId": null
}