Saltar al contenido

Generación Viral de Títulos y Miniaturas YouTube

Genera títulos y miniaturas virales para YouTube. Inicia con una idea, scrapea YouTube (Apify), calcula el CTR, analiza patrones con LLMs (Mistral, Llama/OpenAI/OpenRouter) y Hugging Face para crear 5 títulos y un prompt de miniatura.

AI 41 nodos 16 tipos conectado
Cargando workflow...

Nodos

ManualTrigger Wait OutputParserStructured ChainLlm HttpRequest LmChatMistralCloud If Code Merge StickyNote Set Html SplitInBatches OpenAi ExtractFromFile Filter

Herramientas

Apify Mistral Cloud OpenAI OpenRouter Hugging Face

Detalles

ID
4504
Nodos
41
Conex.
Tipos
16

Pertenece a:

¿Qué hace este workflow?

Este workflow avanzado permite a creadores de contenido y agencias de marketing generar automáticamente títulos y miniaturas altamente virales para sus videos de YouTube. Partiendo de una idea inicial, el sistema utiliza Apify para scrapear YouTube, identificando patrones y tendencias de contenido exitoso. Luego, emplea modelos de lenguaje avanzados como Mistral Cloud, OpenAI o OpenRouter, junto con Hugging Face, para analizar el CTR (Click-Through Rate) de videos similares y comprender los factores de viralidad. El proceso culmina con la creación de 5 opciones de títulos optimizados y un prompt detallado para generar la miniatura perfecta. Este enfoque basado en datos y IA reduce drásticamente el tiempo de ideación y mejora significativamente las probabilidades de éxito de cada video. Ahorra horas de investigación manual y aumenta la visibilidad y el engagement de tu canal de YouTube.

¿Cómo funciona?

Este workflow usa 41 nodos conectados con 16 tipos diferentes: ManualTrigger, Wait, OutputParserStructured, ChainLlm, HttpRequest y 11 más. La estructura está totalmente conectada — listo para importar.

¿Para quién es?

Diseñado para empresas de Agencias de Marketing. Nivel avanzado — recomendado para usuarios experimentados. Alto valor de negocio: automatiza una tarea recurrente con impacto directo.

¿Lo quieres en tu empresa?

Lo implementamos por ti end-to-end: integración, deploy, mantenimiento y soporte. Consultoría B2B con Genai Sapiens.

Hablemos de tu proyecto

¿Quieres aprender a hacerlo?

Sprints de 30 días con companion IA + comunidad. Aprende n8n, automatización y agentes IA desde cero o nivel avanzado.

Ver formación Momentum

Workflows similares

"},"typeVersion":1.2},{"id":"","name":"Loop Over Items","type":"n8n-nodes-base.splitInBatches","position":[2260,520],"parameters":{"options":{}},"typeVersion":3},{"id":"","name":"Analyze Image","type":"@n8n/n8n-nodes-langchain.openAi","position":[2500,540],"parameters":{"text":"=You are an expert in YouTube thumbnail performance. Your job is to analyze why a thumbnail might be performing well and describe it precisely so a designer or AI can later recreate its essence.\n\nYou will receive:\n- The subject of the video : {{ $('Content Idea').first().json.query }}\n- The URL of the thumbnail image : {{ $json.thumbnailUrl }}\n\nYour goal:\n1. Analyze the thumbnail's visual elements (faces, text, colors, emotions, layout, contrast, focus, etc.)\n2. Identify what makes it visually compelling or attention-grabbing\n3. Relate your analysis back to the video’s subject\n4. Return a structured and precise description of the thumbnail that could be reused to generate a similar performing image.\n\nRespond only with the description. Do not add anything else.","modelId":{"__rl":true,"mode":"list","value":"meta-llama/llama-3.2-11b-vision-instruct:free","cachedResultName":"META-LLAMA/LLAMA-3.2-11B-VISION-INSTRUCT:FREE"},"options":{},"resource":"image","simplify":false,"imageUrls":"={{ $json.thumbnailUrl }}","operation":"analyze"},"credentials":{"openAiApi":{"id":"","name":"OpenAi account 2"}},"typeVersion":1.8},{"id":"","name":"Wait3","type":"n8n-nodes-base.wait","position":[2780,720],"webhookId":"","parameters":{},"typeVersion":1.1},{"id":"","name":"Merge","type":"n8n-nodes-base.merge","position":[3580,40],"parameters":{},"typeVersion":3},{"id":"","name":"Mistral Cloud Chat Model3","type":"@n8n/n8n-nodes-langchain.lmChatMistralCloud","position":[2680,460],"parameters":{"model":"mistral-large-latest","options":{}},"credentials":{"mistralCloudApi":{"id":"","name":"Mistral Cloud account"}},"typeVersion":1},{"id":"","name":"Create Prompt","type":"@n8n/n8n-nodes-langchain.chainLlm","position":[2620,280],"parameters":{"text":"You are an expert visual prompt engineer.","messages":{"messageValues":[{"type":"HumanMessagePromptTemplate","message":"=You are an expert visual prompt engineer.\n\nYou are given several thumbnail analysis texts, each describing what made a YouTube thumbnail visually engaging and effective. \n\nData for you : \n- The Content Idea : {{ $('Content Idea').first().json.query }}\n- The Analysis of performing Thumbnails for this idea : {{ $json.analysis}}\n\n\nYour job is to:\n1. **Summarize the most common and impactful visual elements** across all analyses.\n2. **Generate one precise image prompt** that captures the best-performing elements, ready to be used with an image generation model (like DALL·E, Midjourney, or Stable Diffusion).\nMandatory : \n- Maximum 400 characters (including spaces and punctuations) !\n- When generating the prompt text, make sure that any double quotes within the string are escaped using a backslash (\\\") so that the final output can be safely included in a JSON object.\n\n📥 Input: A list of thumbnail descriptions and visual analyses.\n🎯 Goal: One image prompt that includes clear guidance on:\n- Subject placement and expression\n- Colors and contrast\n- Text style or layout (if any)\n- Graphic or visual elements\n- Overall layout or composition\n- Mood or tone (e.g. excitement, curiosity, professionalism)\n\n⚠️ Don't describe what you’re doing. Just output the final image prompt.\n\n📌 Example output format:\nA high-contrast YouTube thumbnail with a confident man on the right side, bold yellow text on the left saying \"AI SECRETS\", a dark tech-themed background, and digital data stream effects in the back — cinematic lighting, 16:9 composition."}]},"promptType":"define"},"typeVersion":1.5},{"id":"","name":"Create List of Analysis","type":"n8n-nodes-base.code","position":[2440,280],"parameters":{"jsCode":"const analysis = items.map(item => item.json.choices[0].message.content);\n\nreturn [\n {\n json: {\n analysis\n }\n }\n];\n"},"typeVersion":2},{"id":"","name":"Generate Image","type":"n8n-nodes-base.httpRequest","position":[3120,280],"parameters":{"url":"https://api-inference.huggingface.co/models/black-forest-labs/FLUX.1-dev","method":"POST","options":{"response":{"response":{"responseFormat":"file"}}},"jsonBody":"={\n \"inputs\":\"{{ $json.text }}\",\n \"parameters\": {\n \"height\": 720,\n \"width\": 1280,\n \"num_inference_steps\": 30,\n \"guidance_scale\": 8.0\n }\n}\n","sendBody":true,"sendHeaders":true,"specifyBody":"json","headerParameters":{"parameters":[{"name":"Authorization","value":"Bearer [YOUR_API_TOKEN]"},{"name":"Accept","value":"image/png"},{"name":"Content-Type","value":"application/json"}]}},"typeVersion":4.2},{"id":"","name":"Convert to Base64","type":"n8n-nodes-base.extractFromFile","position":[3320,280],"parameters":{"options":{"encoding":"base64"},"operation":"binaryToPropery"},"typeVersion":1},{"id":"","name":"Sticky Note3","type":"n8n-nodes-base.stickyNote","position":[-760,-100],"parameters":{"color":7,"width":460,"height":360,"content":"## 1- Input\nEnter your content idea in the Edit Fields node in a \"raw\" format. Ex : Boil Eggs Perfectly"},"typeVersion":1},{"id":"","name":"Sticky Note4","type":"n8n-nodes-base.stickyNote","position":[-280,-100],"parameters":{"color":7,"width":820,"height":480,"content":"## 2- Create DataSet\nLLM create 3 keywords request based on the idea and Apify scrape the YTB Search"},"typeVersion":1},{"id":"","name":"Sticky Note5","type":"n8n-nodes-base.stickyNote","position":[560,-100],"parameters":{"color":7,"width":680,"height":520,"content":"## 3 - Wait for DataSet Completion\nWait until the dataset is completed in Apify"},"typeVersion":1},{"id":"","name":"Sticky Note6","type":"n8n-nodes-base.stickyNote","position":[1260,-100],"parameters":{"color":7,"width":860,"height":460,"content":"## 4- Filter Performing Videos\nRetrieve Dataset from Apify, calculate approximation of CTR and filter top performing videos"},"typeVersion":1},{"id":"","name":"Sticky Note7","type":"n8n-nodes-base.stickyNote","position":[2140,-340],"parameters":{"color":7,"width":1340,"height":500,"content":"## 5- Generate Titles\nLLM analyze patterns of best performing titles and create a prompt based on it. Another LLM create 5 titles based on these criteria"},"typeVersion":1},{"id":"","name":"Sticky Note8","type":"n8n-nodes-base.stickyNote","position":[2140,180],"parameters":{"color":7,"width":1340,"height":760,"content":"## 6- Generate Thumbnail\nLLM analyze patterns of best performing thumbnails and create a prompt based on it. Another LLM create 1 thumbnail based on these criteria"},"typeVersion":1},{"id":"","name":"Sticky Note9","type":"n8n-nodes-base.stickyNote","position":[3500,-100],"parameters":{"color":7,"width":520,"height":360,"content":"## 7- Output\nReturn titles and thumbnail in a HTML Page"},"typeVersion":1},{"id":"","name":"Keep Top Performing Videos","type":"n8n-nodes-base.filter","position":[1900,20],"parameters":{"options":{},"conditions":{"options":{"version":2,"leftValue":"","caseSensitive":true,"typeValidation":"strict"},"combinator":"and","conditions":[{"id":"","operator":{"type":"number","operation":"gt"},"leftValue":"={{ $json.ctrPercent }}","rightValue":0.1}]}},"typeVersion":2.2},{"id":"","name":"Analyze Pattern & Create Prompt","type":"@n8n/n8n-nodes-langchain.chainLlm","position":[2600,-240],"parameters":{"text":"=You are an expert in YouTube video title optimization. Here is your input : \n{{ $json.titles.join('\\n') }}","messages":{"messageValues":[{"type":"HumanMessagePromptTemplate","message":"=You will be given a list of YouTube video titles that ranked highly for a specific keyword or topic. For information the topic of the video is : {{ $('Content Idea').first().json.query }}\nYour task is to: \n1. Analyze the title list for patterns, such as: \n- Repeating structures or formats (e.g., “How to…”, “Top X…”, “This Changed Everything”) \n- Use of numbers, brackets, capital letters, emotional hooks, urgency, etc. \n- Word combinations or topic angles (e.g., “AI Automation for Creators” vs. “Build AI Tools with No Code”) \n2. Summarize what makes these titles effective (mandatory : be as precise as possible in 5-15 points). \n3. Based on your analysis, generate a prompt for another AI language model. This prompt will instruct the second model to generate new YouTube titles for the same keyword/topic, using the same high-performing structure, tone, and techniques. \n\n⚠️ Make sure the new prompt: \n- Clearly instructs the LLM to follow identified title patterns. - Tells it to generate clickable, short, high-performing YouTube titles. \n- Avoids common spammy phrasing or clickbait that doesn’t match the original tone. \n\n✅ Format your response like this:\nPATTERN SUMMARY:\n- Point 1\n- Point 2\n- Point 3\n\nPROMPT:\n\"Create 5 YouTube video titles about [TOPIC] using the following format:\n- [Insert pattern or tone summary]\n- Each title should be under 60 characters, engaging, and based on the provided structure.\"\n\n📥 Example Input (titles):\n- 5 AI Tools That Will Replace Your Job in 2025\n- The Best No-Code AI Tool You’re Not Using Yet\n- I Built an AI Business with $0 (Full Tutorial)\n- How to Automate Everything with AI Agents\n- This Free AI Tool Blew My Mind\n\n🧠 Expected Output:\nPATTERN SUMMARY:\n- Uses emotional triggers like fear or curiosity (“Replace Your Job”, “Blew My Mind”)\n- Often includes numbers or power words (“5 Tools”, “Best”, “Free”)\n- Many titles promise results or value (“Full Tutorial”, “Automate Everything”)\n\nPROMPT:\n\"Create 5 YouTube video titles about AI automation tools for solo creators using this format:\n- Include emotional or curiosity-based phrases\n- Use numbers or power words when relevant\n- Promise a clear value or benefit in the title\nEach title should be under 60 characters and match the tone of the top-ranking videos.\"\n"}]},"promptType":"define"},"typeVersion":1.5},{"id":"","name":"Wait4","type":"n8n-nodes-base.wait","position":[2940,280],"webhookId":"","parameters":{},"typeVersion":1.1}],"active":false,"pinData":{},"settings":{"executionOrder":"v1"},"versionId":"","connections":{"If":{"main":[[{"node":"Get DataSet","type":"main","index":0}],[{"node":"Wait2","type":"main","index":0}]]},"Wait":{"main":[[{"node":"Create Titles","type":"main","index":0}]]},"Merge":{"main":[[{"node":"HTML","type":"main","index":0}]]},"Wait1":{"main":[[{"node":"YTB Search Scrape","type":"main","index":0}]]},"Wait2":{"main":[[{"node":"Check IF Finished","type":"main","index":0}]]},"Wait3":{"main":[[{"node":"Loop Over Items","type":"main","index":0}]]},"Wait4":{"main":[[{"node":"Generate Image","type":"main","index":0}]]},"Merge Data":{"main":[[{"node":"Keep Top Performing Videos","type":"main","index":0}]]},"Get DataSet":{"main":[[{"node":"Calculate CTR","type":"main","index":0},{"node":"Merge Data","type":"main","index":0}]]},"Content Idea":{"main":[[{"node":"Create Keywords","type":"main","index":0}]]},"Analyze Image":{"main":[[{"node":"Wait3","type":"main","index":0}]]},"Calculate CTR":{"main":[[{"node":"Merge Data","type":"main","index":1}]]},"Create Prompt":{"main":[[{"node":"Wait4","type":"main","index":0}]]},"Create Titles":{"main":[[{"node":"Merge","type":"main","index":0}]]},"Generate Image":{"main":[[{"node":"Convert to Base64","type":"main","index":0}]]},"Create Keywords":{"main":[[{"node":"Wait1","type":"main","index":0}]]},"Loop Over Items":{"main":[[{"node":"Create List of Analysis","type":"main","index":0}],[{"node":"Analyze Image","type":"main","index":0}]]},"Check IF Finished":{"main":[[{"node":"If","type":"main","index":0}]]},"Convert to Base64":{"main":[[{"node":"Merge","type":"main","index":1}]]},"YTB Search Scrape":{"main":[[{"node":"Check IF Finished","type":"main","index":0}]]},"Create List of Titles":{"main":[[{"node":"Analyze Pattern & Create Prompt","type":"main","index":0}]]},"Create List of Analysis":{"main":[[{"node":"Create Prompt","type":"main","index":0}]]},"Mistral Cloud Chat Model":{"ai_languageModel":[[{"node":"Create Keywords","type":"ai_languageModel","index":0}]]},"Structured Output Parser":{"ai_outputParser":[[{"node":"Create Keywords","type":"ai_outputParser","index":0}]]},"Mistral Cloud Chat Model1":{"ai_languageModel":[[{"node":"Analyze Pattern & Create Prompt","type":"ai_languageModel","index":0}]]},"Mistral Cloud Chat Model2":{"ai_languageModel":[[{"node":"Create Titles","type":"ai_languageModel","index":0}]]},"Mistral Cloud Chat Model3":{"ai_languageModel":[[{"node":"Create Prompt","type":"ai_languageModel","index":0}]]},"Structured Output Parser1":{"ai_outputParser":[[{"node":"Create Titles","type":"ai_outputParser","index":0}]]},"Keep Top Performing Videos":{"main":[[{"node":"Loop Over Items","type":"main","index":0},{"node":"Create List of Titles","type":"main","index":0}]]},"Analyze Pattern & Create Prompt":{"main":[[{"node":"Wait","type":"main","index":0}]]},"When clicking ‘Test workflow’":{"main":[[{"node":"Content Idea","type":"main","index":0}]]}}}