使用Dumpling AI和GPT-4o生成YouTube视频创意
中级
这是一个Content Creation, Multimodal AI领域的自动化工作流，包含 13 个节点。主要使用 Code, Wait, Gmail, SplitOut, Aggregate 等节点。 通过视频分析使用Dumpling AI和GPT-4o生成YouTube内容创意
前置要求
- •Google 账号和 Gmail API 凭证
- •可能需要目标 API 的认证凭证
- •Google Sheets API 凭证
- •OpenAI API Key
使用的节点 (13)
工作流预览
可视化展示节点连接关系，支持缩放和平移
导出工作流
复制以下 JSON 配置到 n8n 导入，即可使用此工作流
{
"id": "Iqn6p06vxqQB1o2B",
"meta": {
"instanceId": "a1ae5c8dc6c65e674f9c3947d083abcc749ef2546dff9f4ff01de4d6a36ebfe6",
"templateCredsSetupCompleted": true
},
"name": "使用 Dumpling AI 和 GPT-4o 生成 YouTube 视频创意",
"tags": [],
"nodes": [
{
"id": "dd360394-c997-46eb-b4aa-813ae2efb284",
"name": "新 YouTube 视频行触发",
"type": "n8n-nodes-base.googleSheetsTrigger",
"position": [
-544,
-320
],
"parameters": {
"event": "rowAdded",
"options": {},
"pollTimes": {
"item": [
{
"mode": "everyMinute"
}
]
},
"sheetName": {
"__rl": true,
"mode": "list",
"value": 943292970,
"cachedResultUrl": "https://docs.google.com/spreadsheets/d/1Jq2UpzUq8lBMNJxezZLErnlqH4z7tKI2h2jONNA3lEc/edit#gid=943292970",
"cachedResultName": "YouTube finds"
},
"documentId": {
"__rl": true,
"mode": "list",
"value": "1Jq2UpzUq8lBMNJxezZLErnlqH4z7tKI2h2jONNA3lEc",
"cachedResultUrl": "https://docs.google.com/spreadsheets/d/1Jq2UpzUq8lBMNJxezZLErnlqH4z7tKI2h2jONNA3lEc/edit?usp=drivesdk",
"cachedResultName": "Youtube Video/Creators"
}
},
"credentials": {
"googleSheetsTriggerOAuth2Api": {
"id": "qDzHSzTkclwDHpSR",
"name": "Google Sheets Trigger account"
}
},
"typeVersion": 1
},
{
"id": "2a65182b-3fdb-4525-a761-56e8de94a1c0",
"name": "循环处理视频",
"type": "n8n-nodes-base.splitInBatches",
"position": [
-144,
-320
],
"parameters": {
"options": {}
},
"typeVersion": 3
},
{
"id": "8c9d3978-fad8-4a55-bf71-f25b9981e2d0",
"name": "请求间等待",
"type": "n8n-nodes-base.wait",
"position": [
128,
-176
],
"webhookId": "e8a41fd5-d916-47de-b9a4-def436348c2c",
"parameters": {},
"typeVersion": 1.1
},
{
"id": "0ea80688-8018-4586-8440-d52023c8bd4b",
"name": "从 Dumpling AI 获取字幕",
"type": "n8n-nodes-base.httpRequest",
"position": [
352,
-176
],
"parameters": {
"url": "https://app.dumplingai.com/api/v1/get-youtube-transcript",
"method": "POST",
"options": {},
"jsonBody": "={\n \"videoUrl\": \"{{ $json['Video Link'] }}\"\n} ",
"sendBody": true,
"specifyBody": "json",
"authentication": "genericCredentialType",
"genericAuthType": "httpHeaderAuth"
},
"credentials": {
"httpHeaderAuth": {
"id": "RLFzAcGRepr5eXZB",
"name": "Dumpling AI-n8n"
}
},
"typeVersion": 4.2
},
{
"id": "28d28895-e29b-4631-863f-e41d066582c4",
"name": "从 Dumpling AI 获取评论",
"type": "n8n-nodes-base.httpRequest",
"position": [
576,
-176
],
"parameters": {
"url": "https://app.dumplingai.com/api/v1/youtube/video/comments",
"method": "POST",
"options": {},
"jsonBody": "={\n \"url\": \"{{ $('Trigger on New YouTube Video Row').item.json['Video Link'] }}\"\n}",
"sendBody": true,
"specifyBody": "json",
"authentication": "genericCredentialType",
"genericAuthType": "httpHeaderAuth"
},
"credentials": {
"httpHeaderAuth": {
"id": "RLFzAcGRepr5eXZB",
"name": "Dumpling AI-n8n"
}
},
"typeVersion": 4.2
},
{
"id": "97d55e38-bb35-4451-8bdd-17299cfbab57",
"name": "提取评论内容",
"type": "n8n-nodes-base.code",
"position": [
800,
-176
],
"parameters": {
"jsCode": "// n8n Function node\n// Input: items[0].json.comments (array of comment objects)\n// Output: one item per comment with { content: \"...\" }\n\nconst results = [];\n\nfor (const item of items) {\n if (item.json.comments && Array.isArray(item.json.comments)) {\n for (const comment of item.json.comments) {\n results.push({\n json: {\n content: comment.content\n }\n });\n }\n }\n}\n\nreturn results;\n"
},
"typeVersion": 2
},
{
"id": "1793f720-9f0b-4701-8eb6-2dee3d4d875b",
"name": "合并评论到单个字段",
"type": "n8n-nodes-base.aggregate",
"position": [
1024,
-176
],
"parameters": {
"include": "specifiedFields",
"options": {},
"aggregate": "aggregateAllItemData",
"fieldsToInclude": "content",
"destinationFieldName": "comment"
},
"typeVersion": 1
},
{
"id": "11577acc-7188-48fe-a0ff-53b5a0a65531",
"name": "使用 GPT-4o 生成视频创意",
"type": "@n8n/n8n-nodes-langchain.openAi",
"position": [
1248,
-176
],
"parameters": {
"modelId": {
"__rl": true,
"mode": "list",
"value": "chatgpt-4o-latest",
"cachedResultName": "CHATGPT-4O-LATEST"
},
"options": {},
"messages": {
"values": [
{
"role": "system",
"content": "=You are a YouTube content strategist. I will provide you with three inputs: 1) the transcript of a YouTube video, 2) the topic or search query that led to this video, and 3) a set of comments viewers left on the video.\n\nYour task:\n\nAnalyze the transcript to identify the main themes, insights, and information shared.\n\nReview the search topic to understand what people are actively looking for.\n\nAnalyze the comments to detect audience sentiment, recurring questions, pain points, and what viewers wish had been covered.\n\nBased on this analysis, generate 3 to 5 concrete YouTube video ideas for my channel. For each idea, clearly explain:\n\nThe proposed video title or topic.\n\nWhy this idea is a good follow-up (tie it directly to the transcript, search intent, or comments).\n\nHow it can attract engagement (views, likes, comments).\n\nMake your response clear, structured, and actionable so I can immediately plan my next video.\n\nReturn the result strictly in JSON format with the following structure:\n\n{\n \"contentIdeas\": [\n {\n \"title\": \"string\",\n \"whyGoodIdea\": \"string\",\n \"engagementPotential\": \"string\"\n }\n ]\n}\n\nMake sure the response is valid JSON only, with no extra text or explanation outside of the JSON."
},
{
"content": "=comment: {{ JSON.stringify($json.comment) }}\n\ntranscript:{{ $('Get Transcript from Dumpling AI').item.json.transcript }}\n\nSearch Topic:{{ $('Trigger on New YouTube Video Row').item.json['search topic'] }}"
}
]
},
"jsonOutput": true
},
"credentials": {
"openAiApi": {
"id": "dd8NvMC6rvx8RITo",
"name": "OpenAi account 2"
}
},
"typeVersion": 1.8
},
{
"id": "152a63c2-39e1-4224-ad00-63d27228b938",
"name": "拆分内容创意",
"type": "n8n-nodes-base.splitOut",
"position": [
1600,
-176
],
"parameters": {
"options": {},
"fieldToSplitOut": "message.content.contentIdeas"
},
"typeVersion": 1
},
{
"id": "b81bf143-f4dc-47db-99c3-76fd77248fa6",
"name": "保存视频创意到 Google Sheets",
"type": "n8n-nodes-base.googleSheets",
"position": [
1824,
-176
],
"parameters": {
"columns": {
"value": {
"title": "={{ $json.title }}",
"whyGoodIdea": "={{ $json.whyGoodIdea }}",
"engagementPotential": "={{ $json.engagementPotential }}"
},
"schema": [
{
"id": "title",
"type": "string",
"display": true,
"required": false,
"displayName": "title",
"defaultMatch": false,
"canBeUsedToMatch": true
},
{
"id": "whyGoodIdea",
"type": "string",
"display": true,
"required": false,
"displayName": "whyGoodIdea",
"defaultMatch": false,
"canBeUsedToMatch": true
},
{
"id": "engagementPotential",
"type": "string",
"display": true,
"required": false,
"displayName": "engagementPotential",
"defaultMatch": false,
"canBeUsedToMatch": true
}
],
"mappingMode": "defineBelow",
"matchingColumns": [],
"attemptToConvertTypes": false,
"convertFieldsToString": false
},
"options": {},
"operation": "append",
"sheetName": {
"__rl": true,
"mode": "list",
"value": "",
"cachedResultUrl": "https://docs.google.com/spreadsheets/d/1Jq2UpzUq8lBMNJxezZLErnlqH4z7tKI2h2jONNA3lEc/edit#gid=1868697318",
"cachedResultName": "Youtube Content Idea"
},
"documentId": {
"__rl": true,
"mode": "list",
"value": "",
"cachedResultUrl": "https://docs.google.com/spreadsheets/d/1Jq2UpzUq8lBMNJxezZLErnlqH4z7tKI2h2jONNA3lEc/edit?usp=drivesdk",
"cachedResultName": "Youtube Video/Creators"
}
},
"credentials": {
"googleSheetsOAuth2Api": {
"id": "GaJqJHuS5mQxap7q",
"name": "Google Sheets account"
}
},
"typeVersion": 4.7
},
{
"id": "927b3dfa-43df-4b77-8bde-83442391cecb",
"name": "邮件发送内容创意",
"type": "n8n-nodes-base.gmail",
"position": [
64,
-416
],
"webhookId": "8cb7d891-5779-486b-9348-de21d355dd5c",
"parameters": {
"sendTo": "example@gmail.com",
"message": "=Hello, \n\nHere are the recommended content ideas generated from the latest YouTube video analysis:\n\nhttps://docs.google.com/spreadsheets/d//edit?usp=sharing\n\nThese insights are based on the video transcript, the original search topic, and what viewers are actively discussing in the comments. You can use them to plan your next upload with confidence. Best regards, Your Automation Workflow",
"options": {
"appendAttribution": false
},
"subject": "New YouTube Content Ideas Based on Video Analysis",
"emailType": "text"
},
"credentials": {
"gmailOAuth2": {
"id": "j70r3RTMED1pgN3R",
"name": "Gmail account 2"
}
},
"typeVersion": 2.1
},
{
"id": "35b3ed75-af8f-4d41-9a77-d10c5cb5d44c",
"name": "便签",
"type": "n8n-nodes-base.stickyNote",
"position": [
-608,
-688
],
"parameters": {
"width": 1664,
"height": 896,
"content": "## 📌 基于 Dumpling AI + GPT-4o 的 YouTube 创意生成"
},
"typeVersion": 1
},
{
"id": "289785ab-275e-4087-a851-99c70f30ef1b",
"name": "便签 1",
"type": "n8n-nodes-base.stickyNote",
"position": [
1152,
-672
],
"parameters": {
"color": 4,
"width": 944,
"height": 832,
"content": "## 🤖 生成并保存视频创意"
},
"typeVersion": 1
}
],
"active": false,
"pinData": {
"Extract Comment Content": [
{
"json": {
"content": "Join the fastest-growing AI education platform! Try it free and explore 20+ top-rated courses in AI: https://bit.ly/futurepediaSL"
}
},
{
"json": {
"content": "New to channel & new to AI, but just came here to say not only do i appreciate how well you explain things, but the humor that comes along with it makes it top tier. Subscriber for life. Thanks!"
}
},
{
"json": {
"content": "Fantastic 80/20 walkthrough! The step-ups—from schedule → OpenWeather → Gmail, to form-trigger → Switch → Gmail/Sheets, and then the AI Agent (brain + memory + tools) with “Call NADEN Workflow” and HTTP Request as the universal adapter—make the platform click fast. Loved the practical bits too: pinning test data, reading executions, using Set/Merge for data hygiene, and the self-hosted vs cloud notes.\nOne takeaway I’d add from productionizing flows: for deterministic business automations, keep temperature/top-p/penalties low and put the leverage on prompt design + testing. Version your prompts, keep a tiny regression suite of representative cases, and do quick A/B runs across target models after model updates—drift shows up fast. Also helpful: enforce JSON/structured outputs with validation before downstream calls, add timeouts/retries and idempotency on external actions (email/calendar), and centralize secrets/envs.\nThis was a masterclass in reducing “weeks of alignment” to “minutes of working demos.” Bookmarked and shared—thank you!"
}
},
{
"json": {
"content": "Well done. Clear, simple examples that really showcase the platform’s capabilities."
}
},
{
"json": {
"content": "Amazing! Your style of explaining things makes it easier for me to grasp. Thank you! Keep it up!"
}
},
{
"json": {
"content": "Best video you have made! I learned a lot from this. Keep making stuff like this, total walk through with simple steps and simple explanations. Thank you."
}
},
{
"json": {
"content": "The best n8n introduction ever!"
}
},
{
"json": {
"content": "Awesome content! Thank you so much!"
}
},
{
"json": {
"content": "This is fantastic. The future is here now."
}
},
{
"json": {
"content": "Sorry if you covered in another video, but are you using an AI tool for your motion graphics? Very high production value."
}
},
{
"json": {
"content": "I learned something today. Thank you"
}
},
{
"json": {
"content": "Have you called a sub-workflow from within a loop? For the iteration, sometimes it doesn't return it as an array to the calling workflow. Also you have balls publishing videos without QAing them. That is confidence."
}
},
{
"json": {
"content": "Great video!"
}
},
{
"json": {
"content": "I loved this video! anyway we can get one where you take finished workflows and integrate them into websites? For example we made the fill form in this video, how would i put that fill form on their website?"
}
},
{
"json": {
"content": "Thank you. Excellent video. Please can you upload in 4K in future too"
}
},
{
"json": {
"content": "Great teaching! For the self-hosting by Hostinger, is it necceasy to get the daily auto-backup plan?"
}
},
{
"json": {
"content": "앞으로도\n좋은 영상 \n많이 올려주세요😊"
}
},
{
"json": {
"content": "Great way to learn the key features of n8n. Just a quick question from a novice - if I am to design a workflow for my client, how does the authentication stuff work? Do we have to ask clients to do a google sign-in n8n nodes after creating their own n8n account?"
}
},
{
"json": {
"content": "Great video thanks"
}
},
{
"json": {
"content": "You outdo yourself every day."
}
}
],
"Get Comments from Dumpling AI": [
{
"json": {
"comments": [
{
"id": "Ugze-MG9dX6BZGiQGZV4AaABAg",
"author": {
"name": "@futurepedia_io",
"avatarUrl": "https://yt3.ggpht.com/7gpWN95XOZ-MO30txO1hyDUM9b830cX3mMANqWNVLVegO4LaoWueslV370g-8dBdkirYOKhUCw=s88-c-k-c0x00ffffff-no-rj",
"channelId": "UC_RovKmk0OCbuZjA8f08opw",
"isCreator": true,
"channelUrl": "https://youtube.com/@futurepedia_io",
"isVerified": true
},
"content": "Join the fastest-growing AI education platform! Try it free and explore 20+ top-rated courses in AI: https://bit.ly/futurepediaSL",
"engagement": {
"likes": 4,
"replies": 3
},
"replyLevel": 0,
"publishedTime": "2025-09-03T23:34:04.895Z",
"publishedTimeText": "2 weeks ago"
},
{
"id": "UgwGfJZeK8LVvC11Nzl4AaABAg",
"author": {
"name": "@TylerMayfield-v6j",
"avatarUrl": "https://yt3.ggpht.com/bkDhSdmkTfxEtM78kDw7kOFD3JfMMWwme1Lhi_WBur8aM-NwYaJ1ZykOCyVeHRmaQj0htD5X=s88-c-k-c0x00ffffff-no-rj",
"channelId": "UCXVtt-Z9Pf2wqtzv84Y0JaQ",
"isCreator": false,
"channelUrl": "https://youtube.com/@TylerMayfield-v6j",
"isVerified": false
},
"content": "New to channel & new to AI, but just came here to say not only do i appreciate how well you explain things, but the humor that comes along with it makes it top tier. Subscriber for life. Thanks!",
"engagement": {
"likes": 2,
"replies": 0
},
"replyLevel": 0,
"publishedTime": "2025-09-05T23:34:04.895Z",
"publishedTimeText": "12 days ago"
},
{
"id": "UgwJPjHbh-mwVLeJPxl4AaABAg",
"author": {
"name": "@ДимаГиря-р9д",
"avatarUrl": "https://yt3.ggpht.com/ytc/AIdro_kGjHbZQsmO4n6qoTlnHS277PGHgTA4pUd1TwuCMM2Lhdir-EeXtfYE-3ZZvNWE675VbFo=s88-c-k-c0x00ffffff-no-rj",
"channelId": "UCrA0uYyGcaodo_zgiJ8LixQ",
"isCreator": false,
"channelUrl": "https://youtube.com/@%D0%94%D0%B8%D0%BC%D0%B0%D0%93%D0%B8%D1%80%D1%8F-%D1%809%D0%B4",
"isVerified": false
},
"content": "Fantastic 80/20 walkthrough! The step-ups—from schedule → OpenWeather → Gmail, to form-trigger → Switch → Gmail/Sheets, and then the AI Agent (brain + memory + tools) with “Call NADEN Workflow” and HTTP Request as the universal adapter—make the platform click fast. Loved the practical bits too: pinning test data, reading executions, using Set/Merge for data hygiene, and the self-hosted vs cloud notes.\nOne takeaway I’d add from productionizing flows: for deterministic business automations, keep temperature/top-p/penalties low and put the leverage on prompt design + testing. Version your prompts, keep a tiny regression suite of representative cases, and do quick A/B runs across target models after model updates—drift shows up fast. Also helpful: enforce JSON/structured outputs with validation before downstream calls, add timeouts/retries and idempotency on external actions (email/calendar), and centralize secrets/envs.\nThis was a masterclass in reducing “weeks of alignment” to “minutes of working demos.” Bookmarked and shared—thank you!",
"engagement": {
"likes": 14,
"replies": 3
},
"replyLevel": 0,
"publishedTime": "2025-09-03T23:34:04.895Z",
"publishedTimeText": "2 weeks ago"
},
{
"id": "UgyI0E65gaR-hrah3l94AaABAg",
"author": {
"name": "@shamaribenton",
"avatarUrl": "https://yt3.ggpht.com/ytc/AIdro_lqwAgLag9XqwU4HCDKAJ174FLSOX-0quvwFO1vOCzxwGM=s88-c-k-c0x00ffffff-no-rj",
"channelId": "UCmtICUjrQeSKg_O4605901A",
"isCreator": false,
"channelUrl": "https://youtube.com/@shamaribenton",
"isVerified": false
},
"content": "Well done. Clear, simple examples that really showcase the platform’s capabilities.",
"engagement": {
"likes": 2,
"replies": 0
},
"replyLevel": 0,
"publishedTime": "2025-09-05T23:34:04.895Z",
"publishedTimeText": "12 days ago"
},
{
"id": "Ugxdzuq1HIrpmYB5lpN4AaABAg",
"author": {
"name": "@mutiuodeyemi4152",
"avatarUrl": "https://yt3.ggpht.com/ytc/AIdro_kmCMkj7UC_oGngGoDA7jk4mr8glQmNfJEQtg3KobsQR7j5=s88-c-k-c0x00ffffff-no-rj",
"channelId": "UC_9TF6tXnUnvFMXgQaSLuCA",
"isCreator": false,
"channelUrl": "https://youtube.com/@mutiuodeyemi4152",
"isVerified": false
},
"content": "Amazing! Your style of explaining things makes it easier for me to grasp. Thank you! Keep it up!",
"engagement": {
"likes": 1,
"replies": 0
},
"replyLevel": 0,
"publishedTime": "2025-09-08T23:34:04.895Z",
"publishedTimeText": "9 days ago"
},
{
"id": "UgxKPT9lbZ753TJI1MR4AaABAg",
"author": {
"name": "@rob5800can",
"avatarUrl": "https://yt3.ggpht.com/5-TOzPkK1g54isnuUXCG4bLIwSnKvQHLIVjdKDeHDiIIlENgMPM6r3NYfhrO7skZyB4_AxdcRg=s88-c-k-c0x00ffffff-no-rj",
"channelId": "UCqwhHe9wcrJM95R18zsjUEg",
"isCreator": false,
"channelUrl": "https://youtube.com/@rob5800can",
"isVerified": false
},
"content": "Best video you have made! I learned a lot from this. Keep making stuff like this, total walk through with simple steps and simple explanations. Thank you.",
"engagement": {
"likes": 0,
"replies": 0
},
"replyLevel": 0,
"publishedTime": "2025-09-12T23:34:04.895Z",
"publishedTimeText": "5 days ago"
},
{
"id": "Ugz-7ht-Mv_uOTKk76t4AaABAg",
"author": {
"name": "@jeromefenyo895",
"avatarUrl": "https://yt3.ggpht.com/ytc/AIdro_mszK8nubbgivjJBn-qbdgW_m8T9Oug7cdKYYyftRUPAU4=s88-c-k-c0x00ffffff-no-rj",
"channelId": "UCLmgOAYiPMt3JyswC8VgZvQ",
"isCreator": false,
"channelUrl": "https://youtube.com/@jeromefenyo895",
"isVerified": false
},
"content": "The best n8n introduction ever!",
"engagement": {
"likes": 1,
"replies": 0
},
"replyLevel": 0,
"publishedTime": "2025-09-06T23:34:04.895Z",
"publishedTimeText": "11 days ago"
},
{
"id": "UgxJ6pYpv-w9WlztcA94AaABAg",
"author": {
"name": "@yourlifeonpower",
"avatarUrl": "https://yt3.ggpht.com/ytc/AIdro_kgOmYqepfI5botuM9tlrVwDOQvV8nG5EvRX30oop4_EagaiOjDXA23kwp6xzdzxaokSQ=s88-c-k-c0x00ffffff-no-rj",
"channelId": "UCJau7AtGCXWFWRzUXkWU0fA",
"isCreator": false,
"channelUrl": "https://youtube.com/@yourlifeonpower",
"isVerified": false
},
"content": "Awesome content! Thank you so much!",
"engagement": {
"likes": 1,
"replies": 0
},
"replyLevel": 0,
"publishedTime": "2025-09-11T23:34:04.895Z",
"publishedTimeText": "6 days ago"
},
{
"id": "UgzIPViCaM5h1RvGdFJ4AaABAg",
"author": {
"name": "@VITORB82",
"avatarUrl": "https://yt3.ggpht.com/ytc/AIdro_lA8m1IbeiThCoCW-kC-Wc_lVbTDFSI33Aa-1eGGo1zpQ=s88-c-k-c0x00ffffff-no-rj",
"channelId": "UCo22Pp_89gETJ6OGaOIqy8w",
"isCreator": false,
"channelUrl": "https://youtube.com/@VITORB82",
"isVerified": false
},
"content": "This is fantastic. The future is here now.",
"engagement": {
"likes": 1,
"replies": 0
},
"replyLevel": 0,
"publishedTime": "2025-09-10T23:34:04.895Z",
"publishedTimeText": "7 days ago"
},
{
"id": "UgxzhQL_5m9-hlIvvHx4AaABAg",
"author": {
"name": "@rexyooper1",
"avatarUrl": "https://yt3.ggpht.com/ytc/AIdro_mFL8erclF47nWCTRHqV1Ydd9Gjh9w49efOrfTb7ajXTxA=s88-c-k-c0x00ffffff-no-rj",
"channelId": "UClqz_N5kRL-HNb3zXMfz2wA",
"isCreator": false,
"channelUrl": "https://youtube.com/@rexyooper1",
"isVerified": false
},
"content": "Sorry if you covered in another video, but are you using an AI tool for your motion graphics? Very high production value.",
"engagement": {
"likes": 2,
"replies": 0
},
"replyLevel": 0,
"publishedTime": "2025-09-04T23:34:04.895Z",
"publishedTimeText": "13 days ago"
},
{
"id": "Ugyunszr6WryNd09FNJ4AaABAg",
"author": {
"name": "@meraakifilms",
"avatarUrl": "https://yt3.ggpht.com/VLkkc3M6Cte7RjZWp8hlZnz3L7nMPIymu6o7nGREPCvM17v8QNF1Dp9WSLtsC3cbGLynn4Hn=s88-c-k-c0x00ffffff-no-rj",
"channelId": "UCkowX2tZ5QkUKejFa2GBIMQ",
"isCreator": false,
"channelUrl": "https://youtube.com/@meraakifilms",
"isVerified": false
},
"content": "I learned something today. Thank you",
"engagement": {
"likes": 0,
"replies": 0
},
"replyLevel": 0,
"publishedTime": "2025-09-14T23:34:04.895Z",
"publishedTimeText": "3 days ago"
},
{
"id": "UgxujIUfMUYHlNAV5A94AaABAg",
"author": {
"name": "@zk4761",
"avatarUrl": "https://yt3.ggpht.com/ytc/AIdro_mjEPILjz5gpXlz6vhA9ykOgi1C2G3QuHHvkR-KiPc=s88-c-k-c0x00ffffff-no-rj",
"channelId": "UCR2-VWg2nffUb33JqW6R3zQ",
"isCreator": false,
"channelUrl": "https://youtube.com/@zk4761",
"isVerified": false
},
"content": "Have you called a sub-workflow from within a loop? For the iteration, sometimes it doesn't return it as an array to the calling workflow. Also you have balls publishing videos without QAing them. That is confidence.",
"engagement": {
"likes": 0,
"replies": 0
},
"replyLevel": 0,
"publishedTime": "2025-09-14T23:34:04.895Z",
"publishedTimeText": "3 days ago"
},
{
"id": "UgxSsi0HyVzKDmRv7i14AaABAg",
"author": {
"name": "@txmwa3213",
"avatarUrl": "https://yt3.ggpht.com/-VfGO8R79EpdWuNm8hTWGQJlvr2xH62stpVJkSLVCf6BME91ehLmUnUfYd0_ucQfj2EyW-eRMg=s88-c-k-c0x00ffffff-no-rj",
"channelId": "UCaEUwgQrWeSYr-TJO_MeGjw",
"isCreator": false,
"channelUrl": "https://youtube.com/@txmwa3213",
"isVerified": false
},
"content": "Great video!",
"engagement": {
"likes": 1,
"replies": 0
},
"replyLevel": 0,
"publishedTime": "2025-09-05T23:34:04.895Z",
"publishedTimeText": "12 days ago"
},
{
"id": "Ugw8UWJ-hRrjiAKdcVx4AaABAg",
"author": {
"name": "@RemyB-uz6vy",
"avatarUrl": "https://yt3.ggpht.com/ytc/AIdro_lAyCmPuPTcTtYpOA4QnUiNiel5MSi2UOypDgNoiTtaAGtGVrLt1wtS6foazk69LW0x-w=s88-c-k-c0x00ffffff-no-rj",
"channelId": "UCY-41dYkXUzFpCEYXrahz5A",
"isCreator": false,
"channelUrl": "https://youtube.com/@RemyB-uz6vy",
"isVerified": false
},
"content": "I loved this video! anyway we can get one where you take finished workflows and integrate them into websites? For example we made the fill form in this video, how would i put that fill form on their website?",
"engagement": {
"likes": 0,
"replies": 0
},
"replyLevel": 0,
"publishedTime": "2025-09-09T23:34:04.895Z",
"publishedTimeText": "8 days ago"
},
{
"id": "UgyBwRAJR__pxPQyYrR4AaABAg",
"author": {
"name": "@HeatSeekingMouse",
"avatarUrl": "https://yt3.ggpht.com/ytc/AIdro_kRDhVFOWaHWXMRq94qmyuu6AeabZCvmwIrV6squZIY5Q=s88-c-k-c0x00ffffff-no-rj",
"channelId": "UCILUZX0qGvShFgnxAAhgKmg",
"isCreator": false,
"channelUrl": "https://youtube.com/@HeatSeekingMouse",
"isVerified": false
},
"content": "Thank you. Excellent video. Please can you upload in 4K in future too",
"engagement": {
"likes": 1,
"replies": 0
},
"replyLevel": 0,
"publishedTime": "2025-09-03T23:34:04.895Z",
"publishedTimeText": "2 weeks ago"
},
{
"id": "UgyhT0Dm6M44WLLei-p4AaABAg",
"author": {
"name": "@iamtjah",
"avatarUrl": "https://yt3.ggpht.com/cQPaP02W81le57A0ZCa72AevBsA1XU1XElYKOkY4-g4ZmnQ6vafomXmEQAcSR0UUi6beJBzDvQ=s88-c-k-c0x00ffffff-no-rj",
"channelId": "UCs708-qR6KWyTODpbIPee7Q",
"isCreator": false,
"channelUrl": "https://youtube.com/@iamtjah",
"isVerified": false
},
"content": "Great teaching! For the self-hosting by Hostinger, is it necceasy to get the daily auto-backup plan?",
"engagement": {
"likes": 0,
"replies": 0
},
"replyLevel": 0,
"publishedTime": "2025-09-09T23:34:04.895Z",
"publishedTimeText": "8 days ago"
},
{
"id": "UgzGGdVdNc4Ds8w_W0x4AaABAg",
"author": {
"name": "@Loppatiin",
"avatarUrl": "https://yt3.ggpht.com/ytc/AIdro_lrW1E4FRLJPAuWCFru3EJ_y2e2phRnwHhzC1a2avZkezg=s88-c-k-c0x00ffffff-no-rj",
"channelId": "UCUR6gpbxJhttFNMYIK4o3bw",
"isCreator": false,
"channelUrl": "https://youtube.com/@Loppatiin",
"isVerified": false
},
"content": "앞으로도\n좋은 영상 \n많이 올려주세요😊",
"engagement": {
"likes": 0,
"replies": 0
},
"replyLevel": 0,
"publishedTime": "2025-09-10T23:34:04.895Z",
"publishedTimeText": "7 days ago"
},
{
"id": "UgymbRW2AzgI1EyYzxB4AaABAg",
"author": {
"name": "@junglebook-80",
"avatarUrl": "https://yt3.ggpht.com/6pk7H_K_lHdMC1hWOjfh1CFZWFSZtmuixaHYOtTHOpeBwuVaSePkbx4HRHN_D_yotTgTpiKYAw=s88-c-k-c0x00ffffff-no-rj",
"channelId": "UCL1TpcQPZ6s9GTgGdUlmEig",
"isCreator": false,
"channelUrl": "https://youtube.com/@junglebook-80",
"isVerified": false
},
"content": "Great way to learn the key features of n8n. Just a quick question from a novice - if I am to design a workflow for my client, how does the authentication stuff work? Do we have to ask clients to do a google sign-in n8n nodes after creating their own n8n account?",
"engagement": {
"likes": 0,
"replies": 0
},
"replyLevel": 0,
"publishedTime": "2025-09-08T23:34:04.895Z",
"publishedTimeText": "9 days ago"
},
{
"id": "UgzPmvGiLxQ2IddGSjR4AaABAg",
"author": {
"name": "@edcurrer8992",
"avatarUrl": "https://yt3.ggpht.com/ytc/AIdro_n0RooH69N-D-VSA1XQnRYnn2RD6JsT_bfoj_YHqK7aSdk=s88-c-k-c0x00ffffff-no-rj",
"channelId": "UCNwHlBdz3CpZRIeo6xwZ07g",
"isCreator": false,
"channelUrl": "https://youtube.com/@edcurrer8992",
"isVerified": false
},
"content": "Great video thanks",
"engagement": {
"likes": 0,
"replies": 0
},
"replyLevel": 0,
"publishedTime": "2025-09-13T23:34:04.895Z",
"publishedTimeText": "4 days ago"
},
{
"id": "Ugw0YeBg6efB2pfuK3B4AaABAg",
"author": {
"name": "@DailyReadMotivationHub",
"avatarUrl": "https://yt3.ggpht.com/lL0mLwN1joD1OKRRRjDzgnNIDpsiZ8MAhWbSVmLZdTkuxeiDEOreJbGRjkZa8i-QLEMsmjLYeA=s88-c-k-c0x00ffffff-no-rj",
"channelId": "UCjBEVjqxxzYpKYon6cpUE4Q",
"isCreator": false,
"channelUrl": "https://youtube.com/@DailyReadMotivationHub",
"isVerified": false
},
"content": "You outdo yourself every day.",
"engagement": {
"likes": 0,
"replies": 0
},
"replyLevel": 0,
"publishedTime": "2025-09-06T23:34:04.895Z",
"publishedTimeText": "11 days ago"
}
],
"continuationToken": "Eg0SC2UzT1YzTG5yUzdvGAYy1QIKqwJnZXRfcmFua2VkX3N0cmVhbXMtLUNxY0JDSUFFRlJlMzBUZ2FuQUVLbHdFSTJGOFFnQVFZQnlLTUFic2lOdXZEbW9VS0xZT2xud1hDNVh2MUxCVFFRS2lTSUZxek1YUjNTTzdZSGtQSU1RSEthX1Bfakw0TFdqVzcwWDIwaEJCVVc5YTlJMVBENWx3UWtqQktydndrVzVCMEVkUjFfRHJwTlF4RTRfc2FJUkpaQTF1aG1mcTBJY2xLNkxBQ0lXcUxaV05wWjhxRFMyU1F5ZlFpa0gyUkpxSm9Od3gtVlk1c2xJaElBTm1Sa3M1cFUtWnEtQW9ERUJRU0JRaUlJQmdBRWdjSWh5QVFBUmdBRWdVSXFDQVlBQklIQ0lVZ0VCUVlBUklGQ0lrZ0dBQSIRIgtlM09WM0xuclM3bzAAeAEoFEIQY29tbWVudHMtc2VjdGlvbg%3D%3D"
}
}
],
"Get Transcript from Dumpling AI": [
{
"json": {
"transcript": "00:00 - Naden looks complicated at first, but it's much easier than you think. In this video, I'll show you how to master the 80% you'll actually use. We'll do that by building three automations from\n00:10 - simple to advanced. We start basic, then build all the way up to a full AI personal assistant, an AI agent that can manage your calendar, email, and documents. And it's fully customizable\n00:21 - with tools that matter to you. In my demo, I add things I use like scraping subreddits for the best posts, summarizing YouTube comments, and even generating shorts and posting them\n00:31 - automatically. All of that is controllable from a single chat interface. Those are relevant to me. You'd swap them out for whatever fits your own work. I teach it this way\n00:39 - because you're learning by building something genuinely useful right away. And since you interact with a personal assistant daily, NAN becomes second nature, so you can get comfortable\n00:48 - quickly before you start creating automations for your business, clients, or friends. First, a quick definition for anyone new here. NAND is a node-based automation\n00:58 - platform that has exploded in popularity. One reason for that is because it makes building AI agents surprisingly intuitive. We'll build both, but here's the difference.\n01:07 - Automations are fixed. They follow a set sequence, A to B to C. Agents are dynamic. They can reason, make decisions, and choose which actions to take based on context. We'll get deeper\n01:17 - into agents later when we build one. For now, just know that node-based platforms like NADN can look intimidating at first, but once you learn a few basics, it clicks fast.\n01:27 - So, what can you actually use NAND for? The short answer is a lot. Personal productivity, like building a personal assistant to handle your emails, calendar, and documents. Marketing and\n01:37 - business. Automate social media posting. manage your CRM, handle customer onboarding, or even run entire sales processes. E-commerce, from syncing product data to processing orders and\n01:48 - handling customer support, data and reporting, pull data from multiple sources, clean it up, and generate reports automatically. Home and lifestyle, connect it to smart home\n01:57 - devices, trigger routines based on your time or location, or build little quality of life automations. AI integrations, this is where it gets really fun. hook up language models,\n02:06 - image tools, or custom APIs so you can do things like text summarization, content creation, or even complex agents. And that is just scratching the surface, but the point is whether you're\n02:16 - running a business, automating personal tasks, or just experimenting with AI, NADN can fit in somewhere. Let's jump into our first build. I'm starting with an extremely simple one\n02:28 - because once you see something working, the setup, node types, and features all make a lot more sense. So, I'll circle back to those right after this build so you'll understand them in context. For\n02:38 - this one, we'll create an automation that runs every day, checks the weather, then emails it to me. It's simple on purpose just to show you the core concepts. The next two builds will get\n02:48 - more advanced and much more useful. This is a new workflow. I have a completely blank canvas that I named daily weather. Now, I will click the plus button to create my first node. All workflows will\n02:58 - start with a trigger, something to initiate the workflow. For this one, we'll use the schedule trigger. This will run at whatever time interval I set. There's a few drop downs for the\n03:07 - trigger rules. I want the trigger interval to be days, then one day between triggers, then trigger at hour 6 a.m. on the dot, so I'll leave the minutes at zero. Now, this is set to run\n03:18 - every day at 6 a.m. So, I can click back out to the canvas, and the node is right here. That will initiate the workflow. The next thing it needs to do is get the weather. So, I'll click the plus button\n03:28 - to connect the next node. Now, I have some options. I'll cover all these later. The one we want now will be under action in an app. Do something in an app or service like Google Sheets, Telegram,\n03:37 - or Notion. Or, as you can see, when I open up the list, they have hundreds of pre-made integrations to choose from. I'll also teach you how to integrate tools that aren't on the list. But a\n03:46 - huge amount of stuff already is. The one we're looking for is called Open Weather right here. I just wanted to show the list, but I'll click back out. Normally, I would just type weather up here and\n03:55 - find it faster than scrolling through all of that. When I select it, I get a list of actions to choose from. I want return current weather data and that pulls me right into the node settings\n04:04 - which we select in the middle. Then the left side has the input and the right shows the output. Those will populate once I run this, but we'll go through the settings first. So up top is where\n04:13 - we'll connect our first credentials. This is something you'll have to do for each tool you use. You only need to do it once for each. Then those credentials will be saved for anytime you need to\n04:22 - use them in the future in a new workflow. But you'll do this a lot and it's easier in some than others. With Open Weather, it's simple. Just go to their site, openweather.org. Create an\n04:31 - account. Then from the main page, click API keys and the key is right here. Copy that. Go back to NADN and paste that into the access token field. Save and you're all\n04:41 - done. Now that we have our credentials set up, I'll select the options I want. So, current weather, then switch to Imperial instead of the nonsense metric system. Just kidding. But I do actually\n04:52 - kind of prefer Imperial for temperature for daily use. Anyways, then I will choose by zip code and enter a zip code near me. And that's all the settings. So I'll click out. And you can see we have\n05:02 - the new node here connected to the trigger node. Then I can double click to get back in. Now I can click test step and see if this works so far. Looks good. We can see the structure better\n05:11 - now that we have some information. So other than the trigger nodes, all nodes will have this set up. The left side is the input. The data that's coming from the previous node. In this case, that's\n05:20 - the trigger. So, it has the date and time, then settings in the middle, and then the right side is the output, the data leaving this node to head to the next one. For this, we can see all the\n05:29 - different types of weather data it retrieved. There's a few ways to look at this. We have schema, table, and JSON. Depending on the situation, some are more useful or easier to read than\n05:39 - others. Now, I'll click back to the canvas and click save. Try to get into the habit of saving regularly. Now, I will click the plus button to add our final node. This will be a Gmail node.\n05:48 - We could find it under actions or I'll just type Gmail and select it here. And I want to send a message. That opens into the settings. And on the left, we can see that we have the same data that\n05:57 - was the output from the weather node. Now it's the input for this Gmail node. And first up in here, we need to set up credentials. Again, this one is easy. Just click sign in with Google, choose\n06:07 - your account, approve it, then you're good to go. The next setting I need is I'll type in my email for who I want this sent to, then a basic subject, daily weather. Now I can choose text or\n06:17 - HTML for the email type. I just need text for this. So I'll choose that. Next is the message of the email. I'll start with good morning. The current temperature is. Then I want this to give\n06:27 - me the current weather. So it needs to change dynamically based on the information coming in. To do that, we have all of these data fields over here on the left that are coming from the\n06:36 - previous node. I'll look through and find the temperature, which is under main. Then this box that says temp, I can actually just click and drag that right into the message. Then the JSON\n06:46 - shows up right here. Now the number is going to change every day based on what this temperature variable is that day. And right below it shows what that will actually read as currently. Today's\n06:55 - weather is 76.78. Then I could add in and drag over the location which is down at the bottom here. So I drag over name and drop it in there. Today's weather is 76 in Draper\n07:07 - with a temperature range of drag the minimum temperature to maximum temperature. A wind speed of drag that. There we go. Now they show up as JSON, but NAN makes it easy. You drag and drop\n07:20 - and it previews the text right below. You don't have to know what any of this means, but I will go over later how to read it. It is nice to know and can make things faster for complex setups. And\n07:29 - JSON is just a basic data organization system. It's not super complex, although it does look like some sort of confusing code at first. One last thing you'll typically want to do under options is\n07:39 - this append nad attribution. Switch that off. Otherwise, it adds the tagline sent using NAD at the bottom of every email. Now, the first workflow is complete. So, I'll come back to the canvas. I will\n07:51 - click test workflow and it will run through everything start to finish. It's weirdly satisfying to see all these nodes turn green after you create a workflow. All right, it completed\n07:60 - successfully. Now I can check my email. There it is. Everything exactly how I wanted it. Perfect. And before we move on, just a few canvas tips. Sometimes it can get messy when you're doing more\n08:09 - complex workflows. You can always click the broom and it tidies it up. You can zoom in and out and move around. Then this zoom to fit will center you again. An important one is up at the top. We're\n08:19 - on the editor tab. If I click over to executions, this will show all the previous runs of the workflows. And if it succeeded or failed. This is the one I just ran. The one below is when I\n08:28 - tested the open weather node. And I could click copy to editor to go back to the previous version to work from. Then up at the top, I could switch that to active and it will run at 6 am every\n08:37 - day. Of course, I don't need an email with the weather. That's just to give the basic understanding of what we're doing here. Next, I'll break down the different node types and features you'll\n08:44 - use. Then, we'll build an automation that's actually useful and time-saving. After that, we'll push it further with that customizable AI assistant. I want to cover the account setup. There\n08:55 - are a few different ways to set up an aden. The easiest is the cloud version. Just sign up on NADN's site after the free trial is $20 a month. The other option is self-hosting. NADN is source\n09:06 - available. Not technically open source, but you don't need to worry about that distinction here. The key idea is self-hosting gives you more options, more control, and lower costs. There are\n09:15 - two main ways to self-host. You could do a local install with Docker. That's more technical, lots of configurations. A small percentage of the people watching will want to go that route, so I won't\n09:25 - cover it here. That would probably double the length of this video. The other way to self-host is with a virtual private server, a VPS. It's a much easier middle ground. That's what I use\n09:34 - myself. So, it is self-hosted, but it's easier to set up and cheaper than Naden's cloud. I will show you how to do that with Hostinger, today's sponsor. So, they make it by far the easiest VPS\n09:44 - option I've seen. Unlike most providers, they've built a one-click installation template for NAD. So, you skip almost all the setup. On Hostinger, just go up to services, then down under VPS, select\n09:55 - the self-hosted NAD option. Pick a plan. The KVM2 plan is more than enough for everything in this video and beyond. Then make sure to add the coupon code Futuredia.\n10:06 - After that, standard billing and setup process. Then you'll see your dashboard. Since I selected the NAND option, it's right up here. To open it, click manage app. Now set up your NADN account. And\n10:15 - you can skip through this screen, but this one with the free license key, click this button to send the key. Get that key from your email. Go to your plan and click enter activation key and\n10:24 - paste it in. And that's it. you are up and running with your own self-hosted version of NAN. Now, before we build this next workflow, I want to give you an overview of the\n10:34 - core node types we'll be using. I put these into five categories. We covered two of them already in the first build. First is triggers. Every workflow starts with a trigger. We did on a schedule\n10:43 - already, but the next one will be from a form. Then we'll do chat input. There's also a web hook call when executed by another workflow and more. Actions are the big list I showed of pre-built\n10:54 - integrations of apps and services. Everything from the suite of Google products to the weather tool to NASA, notion, air table, Spotify, Stripe. It's a big list. They let you plug in to\n11:03 - other services and perform actions through their APIs without writing code. HTTP request is also an action node, but I list it separately because it's the universal adapter. This lets you connect\n11:14 - to any API when it's not on the pre-built list. The built-in integrations are really just pre-filled HTTP requests. So once you learn this, you can hook into almost anything. And\n11:24 - logic nodes control how data flows. They can filter, branch, merge, or loop, like if yes, do this, if no, do that. Or they could merge data from multiple streams into one. We'll use some of these in the\n11:35 - next build. The AI agent node is what powers the final build. It connects to a language model, adds memory, and lets you attach as many tools as you want. You give it goals and guidance, and it\n11:46 - decides which tools to use based on your input and its own reasoning. That's my simplified grouping. NAND organizes them a little bit differently, but this structure makes it easier to see how\n11:56 - everything fits together. [Music] Time for something actually useful. We're going to build a form-driven workflow. When someone submits a\n12:05 - sponsorship inquiry, NAD will route it based on the answers, respond to them by email automatically, then log it to a Google sheet. I'm using sponsorship intake as an example because it's\n12:15 - relevant to me, but the pattern works for a ton of use cases. Lead gen, sales funnels, support requests. Basically, anytime a form is filled out, you can trigger a smart sequence of actions and\n12:26 - branch the path based on rules you've set. So, I will create a new workflow, then add a trigger. I'll select form submission. I'm going to use their default test forms. That's for demoing.\n12:36 - Same idea applies no matter what you're using. Like for our actual form and CRM, we use HubSpot. Works with any of this. Now I'll add the form fields I want. So company name, contact name, email,\n12:46 - website, budget. This one I'll switch to number instead of text. And notes I'll switch to text area. I've got this all filled out. Now I can go up and click execute step and it will open up the\n12:58 - form for me to fill out. I will do open AI Sam Altman. I'll use my email so I can see the emails that get sent. Then website for budget. How about $500,000. then notes. People are mad about GPT5.\n13:13 - Save us. Then I'll submit that. Now back in NAND, I see all the data as the output it's sending to the next node. And something very helpful here is this little pin button up here that lets you\n13:23 - pin the data. When I click this, I won't have to fill out this form every time I want to test different parts of this workflow. This data will stay here and send to the next node every time. So\n13:32 - that's a super important button. Next, we're going to set up our first logic node. I'm going to route these forms based on how much money they're offering. so I know if it's worth my\n13:40 - time. Then it will autorespond by email based on that as well. Now this will be under flow then is down here called switch. I'll create some simple rules and for each new rule I add it will\n13:50 - create a separate branch leaving this node. So I'll drag over the budget field and drop it into the first box. Then from the drop down I can choose my options. I want a number that is less\n14:00 - than. Now I can enter a number here 1 million. Then I'll add another rule. Drag budget over. then is greater than or equal to 1 million. That's all I need. Then back on the canvas, you can\n14:14 - see the two outputs leaving this node. The top will be if the sponsor budget is less than 1 million, which is outrageous. Clearly not worth my time. Basically, just an insult. Then the\n14:24 - lower path is a million or more, which is basically the lowest number where I'd maybe consider a shout out. And then based on those completely reasonable numbers, I want to autorespond by email.\n14:35 - So, I'll be adding a Gmail node to each branch. But first, let's actually test this step. All right. Now, we can see that data is routed through only the top path. So, I'll add my first Gmail node\n14:45 - there. I set up a new NAND account for this, so none of my credentials are set up. That way, I'll be sure to not skip any steps you'll see when you first start. And to keep the first build as\n14:53 - simple as possible, I used the NAD cloud version. It is more steps to set up Google credentials when self-hosting. But a very useful feature here, you can click on this documentation link that\n15:03 - links directly to a page that has step-by-step instructions to set that up. Anytime you're stuck on something, you can either click that or asking chatpt is very helpful, too. When you\n15:12 - get errors, you can usually just screenshot them and drop it into chat GPT. Then you'll get a step-by-step fix. Now, I went through the steps in the documentation and got my client ID and\n15:21 - client secret. Now that I've done that, all I'll need to do is just copy and paste that into those same fields for any of the other Google products when I set up the credentials. But now for the\n15:30 - email this will be sent to, I'll drag in the email box from the input. Now that will change automatically to whichever email was submitted on the form. Can do something similar in the subject line.\n15:40 - Hello company. I'll do email type text. Then I need a friendly message. So I'll bring in the name field. And we can see it fills it out below. Sam Altman, how dare you waste my time with this\n15:53 - nonsense. Who do you think you're talking to? Pretty fair response to being lowballed like that and how I respond to most of my emails. But now I will test that step. It was successful.\n16:04 - So I'll check my email. There it is. All right. It feels a lot more real when I actually see an email worded like that. Maybe that's a good negotiation tactic, though. But that did work. Although I\n16:13 - did forget to remove the NAND attribution. So I will go back into the node and turn that off. Now I need to add the Gmail node to the lower route. To save time, I can duplicate the node\n16:22 - up top and drag it down and connect it. Now when I open that up, all the fields are already filled out. I can just change the wording. So this is for a million or over. I'm listening. Anything\n16:33 - you can do to sweeten the deal. I like how the subject line is super bland and then the message is just ridiculous. But now both responses are set up. I could test that by submitting a new form, but\n16:43 - it'll work. I'll test it after I set up the next step which will be adding all this information to a Google sheet. This is an extremely common step used in all sorts of workflows and the same ideas\n16:52 - transfer to other platforms like Air Table, HubSpot, notion, ASA, whatever you might use to store information. So I will add a Google sheet node to the top branch. There's a ton of options for\n17:02 - what you can do with sheets. Append or update a row, create or delete sheets or parts of a sheet. What I want is append row. We'll be adding a new row with all the information from the form that was\n17:12 - submitted. First, we need to add credentials. On Google console, I just need to search for Google Sheets. Click enable. Then, I can paste in the same client ID and client secret from before.\n17:22 - Now, it's going to be updating a document, but I don't have that document yet. So, I need to create one with columns for each of the fields that I'll be adding. So, here's a new sheet. I'll\n17:30 - name it sponsor inquiry tracker. Then add a column for every field that's on the submission form. Then, I'll add one column at the end for action so I know what was done with it. Now, back in\n17:39 - ending, I can find all of this. So under document, I'll paste the URL to the document. Then under sheet, I didn't name this one, so it's just sheet one. Now it shows all the columns that I had\n17:49 - on that sheet. I can drag all this data over from the input section that has every one of these except the final one. For the action on this, I will write decline. That's all I need. I can test\n17:60 - this step. It was successful. So I will go back to the sheet. And there is all the information. Perfect. So now back in NADN, I will duplicate that node and move it to the lower path. Then the only\n18:13 - part I need to change is the decision to follow up sent. This should all be functioning now, but I usually like to rename the nodes so it's more clear and easier to remember what's going on. This\n18:23 - is more important for complex workflows, but I do it on all of them. And you can even come into this switch node to rename the paths. Okay, that looks good. Time to test. So I'll go back and unpin\n18:33 - the data. I'll click clear execution to delete all the data I have in here. Now I'll click test workflow to let it all run in one go. So here's the form this time. How about it's XAI Elon Musk and I\n18:44 - want to test out the other branch. So $5 million for notes. Tell people about all the weird spicy mode stuff we've been adding. Submit form. And then now back here. It looks like it already\n18:56 - completed. That was super fast. It should have sent me an email. There it is. Then I'll check the Google sheet. And it's updated. Perfect. That's a simple version of the workflow. But\n19:05 - before we move on to the agent build, I want to show you some extra ways you could expand or refine this. One option is to enrich the lead. So, for example, the form includes their company name and\n19:15 - website. So, you could add an AI agent node to research the company and generate like a quick summary or if it was in a salesunnel, you might want to, I don't know, pull their LinkedIn\n19:23 - profile or scrape their company site. Then, it could output a short qualification report with selling points or context. That can be super helpful. Another useful one is merge nodes. So\n19:33 - anytime you pull in multiple data streams, you can bring them back together. So for instance, I've built a workflow that scrapes the top posts from six AI subreddits. Then right here, it\n19:42 - merges them all into one data string for analysis and summary. Another one you'll regularly run into is the set node. It lets you clean or reshape your data before passing it on. This pops up in\n19:52 - all sorts of ways, but a simple example, back in the sponsor workflow, I set this budget field to only accept a number. But if I left that as a text box, some people would type in their budget with a\n20:02 - dollar sign. If they did that, the switch node would fail because it's expecting just a number. With a set node, you can strip out symbols or reformat the field so the logic still\n20:11 - works. And those are just a few common patterns. Enrich data, merge streams, clean input. You'll see those in particular come up again and again as you build more advanced workflows.\n20:22 - Now, this is a good time to revisit JSON. You'll see it everywhere in NAND because it's the format data moves through your nodes in. It stands for JavaScript object notation, but all you\n20:32 - really need to know is it's a way to organize information in text. I'll show a few examples in this sponsor workflow. In this switch node where we dragged the budget, the dollar sign JSON just means\n20:43 - it came from the previous node and then it lists the name of the field. So, this just means the budget field from the previous node. Same thing in the email node. All these fields were asking for\n20:53 - information from the previous node. So they all start with dollar sign JSON. Then they state the name of the field. Easy as that. Now in the sheet node, things look a little different. The data\n21:02 - there isn't coming from the last node. It's coming from one earlier in the workflow called on form submission. So we look in that on form submission folder right here. Then find the JSON\n21:12 - item or item.json called company. So then the next one means in the folder on form submission find the JSON item called contact name and so on. And all these extra brackets are used when\n21:25 - something has spaces in it. But I'm not trying to show you how to write all of this, just how to read it. And really that covers the basics. Definitely not everything, just the basics, but it's\n21:34 - not too complicated. There's other types you'll see and different expression patterns. Some can be really useful, like back in this weather workflow. For the temperature, I wouldn't really want\n21:42 - the numbers after the decimal. I just want a clean number. So adding math.round at the beginning then adding round brackets that gives me just the clean number or like I mentioned how I'd\n21:52 - clean the number up in a budget workflow. What that would actually look like is this. So I don't have a ton of these memorized and you don't need to. You can just ask\n22:02 - chatgpt to write them. I'm mentioning it because it's good to know that little changes like this are possible. That way you'll know to ask for them. So again I'll just stress in practice you don't\n22:12 - need to write JSON from scratch. Nadn gives you drag and drop fields and whenever you get stuck, chatbt can explain any JSON snippet or generate new ones for you. It is just helpful to know\n22:22 - the basics so it doesn't look like an alien language when you're using it. And the more workflows you build, the more familiar it becomes. For this next build, we're going to be\n22:31 - using the AI agent node. The automations we've built so far can be incredibly useful, but agents take things to another level. So, here's the difference. Automation equals static and\n22:41 - rulebased. It follows a fixed path from A to B to C. Even if it branches, merges, and uses AI along the way. It's still just executing predefined steps. Agent equals dynamic and flexible. It\n22:52 - can reason, plan, and decide which tools to use based on the context. So, think of it like a digital employee that can think, remember, and get things done. At the core, every agent relies on three\n23:03 - components. There's the brain, the large language model like ChatBT, Claude or Gemini. This is what does the reasoning and language generation. Memory. This gives the agent context. It can remember\n23:14 - past interactions, keep track of a conversation, or pull from external memory sources like documents or a vector database. Then tools are how the agent interacts with the outside world.\n23:24 - Tools can involve retrieving data or context, taking action, or orchestrating other workflows that could also include other agents. Now, NADN's agent node gives you slot for all three of these.\n23:34 - so your agent can think, remember, and act all inside your workflow. One last topic before we jump into the build. HTTP requests and APIs. To go beyond the pre-made integrations, you'll\n23:47 - use an HTTP request node. I mentioned it earlier, but here's a deeper look. So API stands for application programming interface. It's just how software talks to software. So I like the example of a\n23:58 - vending machine. You press a button, that's the request, then the machine gives you something back, the response. You don't need to know what's happening inside. Just give the right input and\n24:07 - you get your result. And the two API requests you'll get most are get, which pulls information like checking the weather, fetching a video, or loading a news feed, and post, which sends\n24:17 - information like submitting a form, adding to a Google sheet, or sending a prompt to chat GBT. There's others, too. Those are the main ones, though. But in short, API is the options like the\n24:28 - buttons on a vending machine. HTTP request is actually pressing one of those buttons. With NAD, you don't have to build everything from scratch. Most things you'll want to connect to are\n24:36 - already there and easy to use, and that's what we've done so far. But you can also build custom tools using HTTP requests to connect to any public API, even when it's not officially\n24:46 - integrated. For a really quick example, back in that simple daily weather automation, let's say I want air quality data. Open Weather doesn't provide it, but air.gov does, though, I check the\n24:56 - list, and there's no built-in integration, so I'd need to use an HTTP request node. This will be a get request. Then we need a URL to put in here. To get that, I'll go to\n25:06 - airnow.gov. You can also just search something like air now api on Google to find it. And from there, I'll create an account and paste in the API code they email me. Once logged in, I go to web\n25:16 - services, current observations by reporting area, and then launch the query tool. I'll enter the same zip code near me, switch the response type to JSON, and click build. that generates a\n25:26 - full URL I can copy, drop it into the URL field of the HTTP request node in NADN. And that's it. Honestly, it's not much harder than a built-in integration. And in this example, what I'd actually\n25:37 - need to do is connect these both to the trigger. Then I could use a merge node to combine the data into one email. But these HTTP requests come up a lot. Each one will have slightly different steps,\n25:46 - but it's usually easy with documentation on their site to follow. Or like always, you can ask Chat GPT to help. We have covered a lot so far and now we're ready for the big one. Building a\n25:58 - personal assistant AI agent. This agent can handle all kinds of tasks, checking your calendar, managing emails, researching topics, even triggering other workflows. And my setup all\n26:08 - connects it to things like summarizing YouTube comments, scraping Reddit posts, and generating short form videos. That's for illustration purpose and also to help spark some ideas of what's\n26:17 - possible. You'd customize it with whatever tools make sense for you. The reason I think this is the best place to start with building agents is you'll actually use it every day. That means\n26:26 - you'll quickly spot mistakes, learn to fix them, and iterate. Just adding tools and refining workflows until it becomes genuinely useful and time-saving. And since you're in full control, it's a\n26:35 - safe place to learn before building agents for clients or your business. Let's get this started. So, like every other workflow, it starts with a trigger. I'll use the chat trigger\n26:43 - inside NAD, so I don't have to keep switching windows. Later, you can swap this out for Slack, WhatsApp, Telegram, Discord, whatever you prefer. Next is the AI agent node. You can find it under\n26:53 - AI or search for it. Now, it has all those slots I mentioned before, the brain, memory, and tools. We'll start with the brain, an LLM of your choice. I'll select OpenAI for this. Inside the\n27:05 - settings, we need credentials. So, with OpenAI, that means we need an API key. Go to platform.openai.com/s. Add funds to your account if you haven't already. This is separate from your\n27:15 - existing Chachi BT subscription. Then click API keys, create new secret key. Give it a name. Choose a project if needed. Then copy the key right away. You won't be able to see it again later.\n27:25 - Paste that into the credentials field in NAN and hit save. Now you can pick a model. I'll select GPT40. That'll work for this. Next, I'll set up the memory. So I'll use their simple memory option.\n27:36 - This is for the built-in chat interface I'm using. If I don't add memory, it forgets everything after every message. So, I'll set this to 10 and I'm good to ask it something. So, I'll click open\n27:45 - chat and ask, "What do you think of NAD?" Now, it will respond using chat GPT. That's working, but it can't do much else for us because we haven't connected\n27:55 - any tools. So, let's add some of the most common for a personal assistant. You'll probably want it to see your calendar. Click the plus button under tools and select Google Calendar. You\n28:04 - can give it varying levels of permission just to read so it can tell what's on your schedule and summarize things. Or you can allow it to make changes. So, you can ask it to add something to your\n28:12 - calendar and it can actually do it. Or you could ask it to move something around and optimize your schedule. Another tool would be Gmail. Same idea for what you'd want it to be able to do.\n28:20 - You can allow it to send emails on your behalf. I'll just let it read for now so it can summarize my emails for me. Then I also want access to a Google sheet. I'll just go with that sponsor intake\n28:29 - sheet we created earlier. Then I'll just use the same document and sheet. Same type of thing we've already done. That's a few tools connected pretty easily. But let's test what we can do already with\n28:38 - just these connected. But before that, the last important step we need to do is add a system prompt. This will be inside the AI agent node. We can give it custom instructions to guide what we want it to\n28:48 - do. The main points to include in your system prompt are the role, in this case, a personal assistant, the instructions, what types of things you want it to do, the tools it has access\n28:58 - to and when to use those tools, then any relevant context about you or your business, and any restrictions. Those won't all apply every time. Just adapt it to what you're working on. Now, the\n29:08 - easiest way to get this is to ask chatbt. Explain what the agent does and what tools it has access to. You may already have an ongoing chat about your build anyway. So, ask it to include\n29:17 - those things. Then, grab that prompt. And to actually add it in here, click add options, then system message, and paste it in. You can see I gave it the role, listed the tools to use, and the\n29:27 - rules for tool usage. That's it. It's good to go. Now, for the test, I'll open up the chat and ask, "Summarize my recent emails." Send that. And looks like it's using the right tool. And\n29:38 - there it is. The recent emails were just the tests from the sponsorship form. But now this will be a pretty good test. I'll ask, do I have any good sponsorship requests? In my system prompt, I asked\n29:48 - it to read the sheet whenever asked about sponsors, not my email. Nice. It used the right tool. And those are the exact requests I entered. Perfect. Now, let's test out the calendar. I'll set up\n29:59 - a call with them. What openings do I have on my calendar tomorrow? Now it has my schedule listed first. Then it identified the gaps I have open to schedule in. Great. And I can say\n30:12 - draft an email response for me and include the two options I have free for meetings during the day. And there it is. That looks good. Although not in my same aggressive tone,\n30:23 - but that'll work. So just like that, connecting a few basic tools, there's already a lot we can do. And when using any of these, like the calendar, email, or sheets, I'll typically want to be\n30:32 - able to do more than read only. I want it to be able to make changes, send emails, add new items, all of that. So, I actually will make a separate workflow like this calendar agent. This can do\n30:41 - all of those things. And the trigger is when executed by another workflow. I do that same thing with Gmail and the other tools. If you try to add all these nodes into one agent, it gets really hard to\n30:50 - stay organized and for the agent to figure out what to do. This makes it much more efficient. But that does bring me to a very powerful aspect of all this. calling other workflows, including\n30:60 - other agents. Using your agent to work with other agents and workflows can add a ton of functionality far beyond just using a calendar to connect to other workflows. It's pretty easy. I can come\n31:10 - back in here and delete the calendar tool. Now, I'll add another tool, but this time I'll use the call naden workflow tool. From there, I choose the calendar agent from the list. Then, it\n31:20 - opens up fields for inputs that I set up in that workflow. If I needed a specific format, I could use that here. or what's best in a lot of cases is clicking these stars which is defined automatically by\n31:30 - the model. Since I'll be interacting using natural language, I can just let the agent know in the system prompt what these fields are for. Then it will adjust how to fill them out based on\n31:38 - whatever I typed. This is what I'd recommend doing after connecting all the main tools that are relevant to you. Build separate workflows that you can test on their own. Then give your agent\n31:47 - access to them. So I'll connect those others that I mentioned and show how this works in practice. Here's the full personal assistant with the calendar, Gmail, and documents as sub aents with\n31:57 - full access to those tools. Then there's a Reddit curator, a YouTube summarizer, and this HTTP request is to a YouTube shorts creator that I demoed in a previous video. Then, of course, I added\n32:07 - all of these tools into the system prompt so it knows what they are and when to call them. Now, I'll open up the chat and ask, "What are the comments like on my recent YouTube video?" And\n32:16 - I'll paste in the link. So, I send that and it uses the brain, then goes straight to using the YouTube summarizer tool. That one works pretty quick. Now, back in the brain to process that. And\n32:26 - there it is. We have the overall impression, sentiment breakdown, helpful critiques, the editing can be overwhelming, parts were unclear, reduced the flashy transitions, and the\n32:36 - positive, it was relatable, made complex AI concepts accessible, graphics and presentation style received praise for being visually appealing. So, the opposite of the critiques. I mean,\n32:46 - that's how YouTube works, though. I'll get comments that say great editing and bad editing, or I go too fast, then others that say I go too slow on the same video. Someone loves it, someone\n32:54 - hates it. That's YouTube. But sometimes there's really helpful critiques mixed in. This also has the top comments, recent comments, content ideas, and mentions the spam. So, I have to like\n33:02 - read through and remove anything that was spam. So, that worked great. Now, how about what are the best recent posts about AI on Reddit? Okay, looks like it's triggered the Reddit curator. This\n33:13 - one takes a minute, so I can jump over to that workflow. Then when I go to executions, it will show that one is currently running. And this is that one I showed before that scrapes six\n33:21 - different AI subreddits. Then it has chatgpt read through all the posts and picks the 10 best. So we'll go back to the agent and looks like that's finishing up going through chat GPT\n33:30 - again. And here's all the posts. It will actually show images on here too if any of them had images as part of the post. Looks like these ones didn't. And all of these are links. I can click on it and\n33:39 - go directly to the posts if it sounds interesting. So perfect. Now I'll say post an AISMR short to my YouTube channel. Here we go. It triggered the right tool and says got it. I've started\n33:50 - the AISMR video workflow. You'll get an email when the video posts. So I'll open up that workflow. Go to executions and we can see this one's running now. But it takes a while since it has to\n33:59 - generate the videos. Then it generates sound effects to go with them. And then it edits it all together. After that it will actually post it, too. When you use the call another workflow tool, your\n34:08 - agent waits for a response. So for this one, I had to set it up as a web hook. That way it just triggers it to run and then responds right away. And this is a workflow I modified from a template.\n34:17 - Naden has thousands of templates you can search through and build from. I made a whole video just showing how to use those templates. This was one of those I demoed. That's why it's using the kind\n34:25 - of outdated AISMR trend. Again, this is just an example. I'll come back when this is finished. But there it is. The full workflow completed. It posted to YouTube, updated a sheet, and sent me an\n34:35 - email. So I can go to YouTube and there it is. Gonna watch a little bit. And there you have it, an AI personal assistant. Hopefully that sparks some ideas for what you can build. I just\n35:02 - want to give a few final tips before we wrap up. So number one is expect errors. They are part of the process. Every workflow is different and even experienced NAD end users hit snags all\n35:12 - the time. The key is to test, iterate, and adjust until it works. It also leads into number two. Use chat GBT as your build partner. There's even a custom GPT trained on the NADN docs that I would\n35:24 - recommend. It's often more helpful than a fresh chat. Number three, keep exploring. There's a ton you can build with just what's in this video. Then a big next step for many people is\n35:33 - retrieval augmented generation that uses a vector database like Pine Cone that lets your agents pull from a much larger knowledge base than just chat history or Google Docs. It opens up a whole new\n35:43 - level of workflows that could include up-to-date and dynamic data from your business. And if you want to go deeper into learning AI, we've built a full course platform on Futuredia with over\n35:53 - 500 lessons across over 20 AI courses. You'll find full learning paths on chat GPT, prompt engineering, custom GPTs, video generation, coding with AI, and a lot more. It's all included in one\n36:05 - subscription. You can get a 7-day free trial using the link in the description. Or a great next step after this video would be watching this one where I walk through exactly how to build from the\n36:15 - free templates on the main NAN site. It's a great way to see how others structure their workflows to get ideas or just use them as"
}
}
],
"Generate Video Ideas with GPT-4o": [
{
"json": {
"index": 0,
"message": {
"role": "assistant",
"content": {
"contentIdeas": [
{
"title": "How to Embed n8n Workflows Into Your Website (With Live Form Example)",
"whyGoodIdea": "Several commenters asked how to take the form shown in the tutorial and embed it directly into a website. This expands on a real user need and advances practical application skills.",
"engagementPotential": "Taps into genuine viewer curiosity, solves a clear pain point, and encourages follow-up comments from users trying to implement the solution on different platforms like WordPress, Webflow, or custom HTML."
},
{
"title": "n8n Authentication Explained: Secure Client Setup for Google, Gmail, and APIs",
"whyGoodIdea": "Multiple novice users asked about how authentication works, especially when building workflows for clients. It's a recurring pain point and one that can create serious roadblocks.",
"engagementPotential": "Solves a major hurdle beginners face; practical videos about OAuth setup, Google authentication, and credential security always generate questions and follow-up comments."
},
{
"title": "From Idea to Execution: Building and Hosting a Custom AI Agent with n8n VPS",
"whyGoodIdea": "The video briefly covered self-hosting via Hostinger, but viewers are interested in deeper productionization workflows (version control, deployment, testing suites, etc.).",
"engagementPotential": "Appeals to more advanced users and those scaling automations for clients; encourages long-term engagement and trust in technical credibility."
},
{
"title": "5 Easy Workflow Templates You Can Copy and Use Today in n8n",
"whyGoodIdea": "The video mentioned templates and there’s active interest from beginners in copying and learning by example. This video could pick 5 practical, ready-to-deploy workflows with walkthroughs.",
"engagementPotential": "Beginner-friendly, immediately actionable content tends to perform well; boosts saves, shares, and comments as users request custom versions of templates."
},
{
"title": "n8n vs Zapier vs Make: Which Automation Tool is Right for You?",
"whyGoodIdea": "While the video sold n8n’s strengths, viewers may be wondering how it compares to other tools they’ve heard of or used. This helps them choose the right fit for their goals.",
"engagementPotential": "Comparison videos generate high engagement and foster discussion in the comments section, especially from users switching tools or sharing experiences."
}
]
},
"refusal": null,
"annotations": []
},
"logprobs": null,
"finish_reason": "stop"
}
}
],
"Trigger on New YouTube Video Row": [
{
"json": {
"Title": "Master 80% of n8n in 36 Minutes",
"Creator": "Futurepedia",
"Video ID": "e3OV3LnrS7o",
"viewCount": "73,068 views",
"Video Link": "https://www.youtube.com/watch?v=e3OV3LnrS7o",
"Video length": "36:23:00",
"search topic": "n8n",
"publishedTimeText": "2 weeks ago"
}
},
{
"json": {
"Title": "You NEED to Use n8n RIGHT NOW!! (Free, Local, Private)",
"Creator": "NetworkChuck",
"Video ID": "ONgECvZNI3o",
"viewCount": "1,629,071 views",
"Video Link": "https://www.youtube.com/watch?v=ONgECvZNI3o",
"Video length": "26:36:00",
"search topic": "n8n",
"publishedTimeText": "2 months ago"
}
},
{
"json": {
"Title": "Master n8n in 2 Hours: Complete Beginner’s Guide for 2025",
"Creator": "Jono Catliff",
"Video ID": "AURnISajubk",
"viewCount": "472,096 views",
"Video Link": "https://www.youtube.com/watch?v=AURnISajubk",
"Video length": "2:10:36",
"search topic": "n8n",
"publishedTimeText": "6 months ago"
}
},
{
"json": {
"Title": "I Tried N8N User Review 2025 - My Honest Experience",
"Creator": "Mike's Guides",
"Video ID": "O1dBnskOg0w",
"viewCount": "6,766 views",
"Video Link": "https://www.youtube.com/watch?v=O1dBnskOg0w",
"Video length": "1:54",
"search topic": "n8n",
"publishedTimeText": "7 months ago"
}
},
{
"json": {
"Title": "Build an AI Email Assistant in n8n | Gmail Automation Walkthrough",
"Creator": "Dave Crawford",
"Video ID": "fNBHxZtNPiY",
"viewCount": "119 views",
"Video Link": "https://www.youtube.com/watch?v=fNBHxZtNPiY",
"Video length": "23:26",
"search topic": "n8n",
"publishedTimeText": "1 day ago"
}
},
{
"json": {
"Title": "N8N vs Flowise (2025) | Which One is Better?",
"Creator": "Bytes Media",
"Video ID": "finbYb8bjL0",
"viewCount": "30,736 views",
"Video Link": "https://www.youtube.com/watch?v=finbYb8bjL0",
"Video length": "2:54",
"search topic": "n8n",
"publishedTimeText": "2 months ago"
}
},
{
"json": {
"Title": "n8n Tutorial for Beginners: How to Build AI Automations for FREE (Step-by-Step)",
"Creator": "Charlie Chang",
"Video ID": "Fy1UCBcgF2o",
"viewCount": "301,187 views",
"Video Link": "https://www.youtube.com/watch?v=Fy1UCBcgF2o",
"Video length": "19:15",
"search topic": "n8n",
"publishedTimeText": "5 months ago"
}
},
{
"json": {
"Title": "Google Opal Is INSANE... Goodbye N8N! (Free Ai App Generator)",
"Creator": "Marcin AI",
"Video ID": "FMTiYeW60Q8",
"viewCount": "121,298 views",
"Video Link": "https://www.youtube.com/watch?v=FMTiYeW60Q8",
"Video length": "11:21",
"search topic": "n8n",
"publishedTimeText": "1 month ago"
}
},
{
"json": {
"Title": "n8n Beginner Course (1/9) - Introduction to Automation",
"Creator": "n8n",
"Video ID": "4BVTkqbn_tY",
"viewCount": "459,078 views",
"Video Link": "https://www.youtube.com/watch?v=4BVTkqbn_tY",
"Video length": "12:48",
"search topic": "n8n",
"publishedTimeText": "1 year ago"
}
},
{
"json": {
"Title": "N8N vs Relevance AI 2025 | Use The Best!",
"Creator": "Mike's Guides",
"Video ID": "lVCeU-swnpA",
"viewCount": "12,237 views",
"Video Link": "https://www.youtube.com/watch?v=lVCeU-swnpA",
"Video length": "1:31",
"search topic": "n8n",
"publishedTimeText": "7 months ago"
}
},
{
"json": {
"Title": "Dont Use n8n... Let Replit Agent 3 Build Your Entire Automations",
"Creator": "AI LABS",
"Video ID": "v1jLYiW0AWc",
"viewCount": "958 views",
"Video Link": "https://www.youtube.com/watch?v=v1jLYiW0AWc",
"Video length": "8:32",
"search topic": "n8n",
"publishedTimeText": "2 hours ago"
}
}
]
},
"settings": {
"executionOrder": "v1"
},
"versionId": "1bf226e2-e083-4eda-bda5-2c3d6f413c8d",
"connections": {
"Loop Over Videos": {
"main": [
[
{
"node": "Email Content Ideas",
"type": "main",
"index": 0
}
],
[
{
"node": "Wait Between Requests",
"type": "main",
"index": 0
}
]
]
},
"Split Content Ideas": {
"main": [
[
{
"node": "Save Video Ideas to Google Sheets",
"type": "main",
"index": 0
}
]
]
},
"Wait Between Requests": {
"main": [
[
{
"node": "Get Transcript from Dumpling AI",
"type": "main",
"index": 0
}
]
]
},
"Extract Comment Content": {
"main": [
[
{
"node": "Merge Comments into Single Field",
"type": "main",
"index": 0
}
]
]
},
"Get Comments from Dumpling AI": {
"main": [
[
{
"node": "Extract Comment Content",
"type": "main",
"index": 0
}
]
]
},
"Get Transcript from Dumpling AI": {
"main": [
[
{
"node": "Get Comments from Dumpling AI",
"type": "main",
"index": 0
}
]
]
},
"Generate Video Ideas with GPT-4o": {
"main": [
[
{
"node": "Split Content Ideas",
"type": "main",
"index": 0
}
]
]
},
"Merge Comments into Single Field": {
"main": [
[
{
"node": "Generate Video Ideas with GPT-4o",
"type": "main",
"index": 0
}
]
]
},
"Trigger on New YouTube Video Row": {
"main": [
[
{
"node": "Loop Over Videos",
"type": "main",
"index": 0
}
]
]
},
"Save Video Ideas to Google Sheets": {
"main": [
[
{
"node": "Loop Over Videos",
"type": "main",
"index": 0
}
]
]
}
}
}常见问题
如何使用这个工作流？
复制上方的 JSON 配置代码，在您的 n8n 实例中创建新工作流并选择「从 JSON 导入」，粘贴配置后根据需要修改凭证设置即可。
这个工作流适合什么场景？
中级 - 内容创作, 多模态 AI
需要付费吗？
本工作流完全免费，您可以直接导入使用。但请注意，工作流中使用的第三方服务（如 OpenAI API）可能需要您自行付费。
相关工作流推荐
使用 Dumpling AI 和 Google 文档的表单转博客自动化
使用 Dumpling AI、OpenAI 和 Google 文档从关键词自动化博客创建
Code
Wait
Limit
+8
17 节点Yang
内容创作
使用 Dumpling AI 从表单自动生成电影级动物视频和声音
使用GPT-4、Dumpling AI和ElevenLabs音频将表单输入转换为电影级视频
Set
Code
Wait
+9
23 节点Yang
内容创作
使用Dumpling AI + GPT-4根据关键词查找TikTok视频问题
使用Dumpling AI + GPT-4根据关键词查找TikTok视频问题
Code
Wait
Limit
+8
16 节点Yang
内容创作
💥 使用NanoBanana、Seedream 4、ChatGPT Image和Veo 3自动化视频广告 - VIDE
使用AI（NanoBanana、Seedream、GPT-4o、Veo 3）自动化和发布视频广告活动
Set
Code
Wait
+16
63 节点Dr. Firas
内容创作
1. 播放列表详情设置机器人副本
使用 Suno、GPT-4、Runway 和 Creatomate 创建 AI 生成的 YouTube 音乐播放列表
If
Set
Code
+22
203 节点Joseph
内容创作
使用Dumpling AI实现Telegram到邮件通讯自动化
使用Dumpling AI和GPT根据Telegram关键词生成邮件通讯
Wait
Gmail
Split Out
+11
16 节点Yang
多模态 AI
