Abdullah Alshiekh This workflow corresponds to n8n.io template #12074 — we link there as the canonical source.
This workflow follows the Agent → Chainsummarization recipe pattern — see all workflows that pair these two integrations.
The workflow JSON
Copy or download the full n8n JSON below. Paste it into a new n8n workflow, add your credentials, activate. Full import guide →
{
"id": "MRINKwuP8BvQjCU1",
"meta": {
"templateCredsSetupCompleted": true
},
"name": "Automate Research with Decodo, OpenAI, and Gmail",
"tags": [],
"nodes": [
{
"id": "0da43c80-0f67-49cc-a23b-b43502b41f83",
"name": "Main Overview",
"type": "n8n-nodes-base.stickyNote",
"position": [
5776,
-1488
],
"parameters": {
"width": 386,
"height": 519,
"content": "### How it works\nThis workflow automates the collection and analysis of AI/LLM research. It runs on a weekly schedule to:\n1. **Search:** Fetch the latest articles using the Decodo API.\n2. **Scrape:** Extract the full text content from each result, filtering out noise.\n3. **Summarize:** Use an LLM to generate dense, technical summaries of every article.\n4. **Report:** An AI Agent analyzes all summaries to score their relevance and compiles a final intelligence report email.\n\n### Setup steps\n1. **Credentials:** Add your `Decodo` and `OpenAI` API keys in the workflow credentials.\n2. **Schedule:** Adjust the **Weekly Trigger** node to your preferred day and time.\n3. **Search Config:** Update the `search_query` in the **Set Search Config** node if you want to track a different topic.\n4. **Email:** Change the recipient email address in both **Send Email** nodes.\n\n### Customization tips\n- Adjust the \"Relevance Score\" prompt in the **Research Analyst Agent** node to make the filtering stricter or looser."
},
"typeVersion": 1
},
{
"id": "62878aa3-dc4d-41d0-bd3b-b05462e19258",
"name": "Section 1",
"type": "n8n-nodes-base.stickyNote",
"position": [
6176,
-1328
],
"parameters": {
"color": 4,
"width": 950,
"height": 256,
"content": "## 1. Search & Filter\nConfigure the search parameters, fetch results from Google via Decodo, and filter out non-article URLs (e.g., YouTube)."
},
"typeVersion": 1
},
{
"id": "389d463e-6da6-4d78-a402-771f11878e7d",
"name": "Section 2",
"type": "n8n-nodes-base.stickyNote",
"position": [
7200,
-1312
],
"parameters": {
"color": 4,
"width": 900,
"height": 592,
"content": "## 2. Scrape & Sanitize\nLoop through each URL, download the raw HTML, and clean it to remove scripts/styles before passing to the LLM."
},
"typeVersion": 1
},
{
"id": "9c87b6e3-05d2-486a-8d74-1552ab66269a",
"name": "Section 3",
"type": "n8n-nodes-base.stickyNote",
"position": [
8176,
-1344
],
"parameters": {
"color": 4,
"width": 480,
"height": 480,
"content": "## 3. Summarize Content\nGenerate concise, dense summaries for each individual article using an LLM Chain."
},
"typeVersion": 1
},
{
"id": "1b10ef37-abe3-4a5a-aa8d-0a4a9f78779d",
"name": "Section 4",
"type": "n8n-nodes-base.stickyNote",
"position": [
8720,
-1344
],
"parameters": {
"color": 4,
"width": 1000,
"height": 480,
"content": "## 4. Analyze & Report\nAn AI Agent reviews all summaries, ranks them by relevance, and sends the final email report."
},
"typeVersion": 1
},
{
"id": "4a88cfde-00ef-466f-80e3-f244241c838f",
"name": "Weekly Trigger",
"type": "n8n-nodes-base.scheduleTrigger",
"position": [
6208,
-1232
],
"parameters": {
"rule": {
"interval": [
{
"daysInterval": 7,
"triggerAtHour": 6
}
]
}
},
"typeVersion": 1.2
},
{
"id": "8467ff7c-d60a-46f1-8daa-94d895dcd0f3",
"name": "Filter & Parse URLs",
"type": "n8n-nodes-base.code",
"position": [
7024,
-1232
],
"parameters": {
"jsCode": "// Safely get the organic results array from Decodo Google Search\n\nconst root = $json.results?.[0]?.content?.results;\n\n// Handle different Decodo response shapes\nconst organic = root?.results?.organic || root?.organic || [];\n\nif (!Array.isArray(organic) || organic.length === 0) {\n return []; // nothing to process\n}\n\n// Filter out YouTube links, then map\nreturn organic\n .filter(item => {\n const url = (item.url || \"\").toLowerCase();\n return !url.includes(\"youtube.com\") && !url.includes(\"youtu.be\");\n })\n .map(item => ({\n json: {\n url: item.url || \"\",\n title: item.title || \"\",\n desc: item.desc || \"\",\n pos: item.pos || item.pos_overall || null,\n url_shown: item.url_shown || \"\",\n favicon_text: item.favicon_text || \"\",\n region: $json.region || \"\",\n platform: $json.platform || \"\"\n }\n }));\n"
},
"typeVersion": 2
},
{
"id": "f0fa2db8-c14c-4e98-925a-324ccb94610b",
"name": "Batch Loop",
"type": "n8n-nodes-base.splitInBatches",
"position": [
7232,
-1232
],
"parameters": {
"options": {}
},
"typeVersion": 3
},
{
"id": "a9f2008f-883a-489f-8934-a162c5ff0166",
"name": "Sanitize HTML",
"type": "n8n-nodes-base.code",
"position": [
7936,
-1120
],
"parameters": {
"mode": "runOnceForEachItem",
"jsCode": "const item = $input.item.json;\n\nlet html = \"\";\n\ntry {\n html = item.results[0].content || \"\";\n} catch (e) {\n html = \"\";\n}\n\nlet cleaned = html\n .replace(/<script[^>]*>[\\s\\S]*?<\\/script>/gi, \"\")\n .replace(/<style[^>]*>[\\s\\S]*?<\\/style>/gi, \"\")\n .replace(/<noscript[^>]*>[\\s\\S]*?<\\/noscript>/gi, \"\")\n .replace(/<\\/?[^>]+>/g, \" \")\n .replace(/ /gi, \" \")\n .replace(/&/gi, \"&\")\n .replace(/"/gi, '\"')\n .replace(/'/gi, \"'\")\n .replace(/</gi, \"<\")\n .replace(/>/gi, \">\")\n .replace(/\\s+/g, \" \")\n .trim();\n\nreturn {\n json: {\n text_clean: cleaned\n }\n};\n"
},
"typeVersion": 2
},
{
"id": "70fe01b3-73ec-41c5-a8cc-f5f512063e59",
"name": "Summarize Chain",
"type": "@n8n/n8n-nodes-langchain.chainSummarization",
"position": [
8208,
-1248
],
"parameters": {
"options": {
"summarizationMethodAndPrompts": {
"values": {
"prompt": "=You will receive cleaned raw text {{ $json.text_clean }} from an AI / LLM related webpage\n(article, blog post, research announcement, release note, or news page).\n\nYour task is to produce a short, dense summary focused ONLY on useful AI / LLM insights.\n\nDo NOT rewrite the entire page.\nDo NOT include marketing fluff, branding, author bios, disclaimers, navigation text, or repeated sections.\n\nExtract ONLY the following information IF it appears in the text:\n\n- Main topic or announcement\n- Model / tool / company name (if any)\n- Type of update (model release, research paper, feature update, benchmark, policy, funding, etc.)\n- Key points or changes (max 3 short bullets)\n- Why it matters (1 short line, practical impact)\n\nReturn the result as a **short human-readable text summary**, NOT JSON.\n\nStyle example:\n\n\u201cTopic: New LLM Model Release \nEntity: Company X \u2013 Model Y \nType: Model release \nKey points: \u2014 ; \u2014 ; \u2014 \nWhy it matters: Faster inference and lower cost for production use.\u201d\n\nIf a field is missing, skip it silently.\nKeep the entire summary under **6\u20138 lines**.\nAvoid extra explanations or speculation.\n",
"summarizationMethod": "stuff"
}
}
}
},
"typeVersion": 2.1
},
{
"id": "9975d7e1-6e4f-434b-b74b-e26f03320f6f",
"name": "Aggregate Summaries",
"type": "n8n-nodes-base.code",
"position": [
8544,
-1248
],
"parameters": {
"jsCode": "// Collect all items coming from Summarization/Extraction\nconst items = $input.all();\n\n// Extract each summary from each item\nconst summaries = items.map(item => item.json);\n\n// Build a single output item\nreturn [\n {\n json: {\n summaries\n }\n }\n];\n"
},
"typeVersion": 2
},
{
"id": "033d4d93-d8f3-4eb1-b843-94fc0bf7138e",
"name": "Research Analyst Agent",
"type": "@n8n/n8n-nodes-langchain.agent",
"position": [
9136,
-1248
],
"parameters": {
"text": "=You are an AI research analyst.\n\nINPUT YOU WILL RECEIVE:\n1. A collection of short summaries generated from AI / LLM related webpages.\n2. Each summary represents one article, announcement, blog post, or research update.\n3. All content comes from the same search context and timeframe.\n\n-----------------------------------\nSTEP 1 \u2014 Understand the Research Focus\n-----------------------------------\nAssume the goal is to track meaningful developments in:\n- AI models and LLMs\n- AI tooling and infrastructure\n- Research breakthroughs\n- Practical production or business impact\n\n-----------------------------------\nSTEP 2 \u2014 Evaluate Each Summary\n-----------------------------------\nFor EACH summary:\n\n1. Determine whether it is **meaningful or actionable**.\n2. Assign a **Relevance Score (0\u2013100%)** based on:\n - Technical significance\n - Novelty (new model, update, result, or capability)\n - Practical or industry impact\n3. Exclude summaries with relevance below **50%**.\n4. For each included summary, prepare:\n - Topic / Title (if clear)\n - Entity (model, company, or organization)\n - Relevance Score (percentage)\n - 1\u20132 short reasons explaining the score\n\n-----------------------------------\nSTEP 3 \u2014 Generate Final Report\n-----------------------------------\nProduce ONE single consolidated report structured as:\n\nA) High-Level Overview \n - 2\u20133 sentences summarizing the overall AI/LLM trend observed.\n\nB) Key Relevant Updates \n - List items ordered by Relevance Score (highest \u2192 lowest):\n - Topic / Entity\n Relevance: XX%\n Why it matters:\n \u2022 <reason 1>\n \u2022 <reason 2>\n\nC) Notable Patterns or Signals \n - 3\u20135 short bullets highlighting:\n \u2022 Repeated themes\n \u2022 Emerging model types or tools\n \u2022 Research vs industry balance\n \u2022 Signals worth monitoring next week\n\n-----------------------------------\nRULES\n-----------------------------------\n- Output must be plain text (no markdown, no JSON).\n- Do NOT invent facts, models, or companies.\n- Do NOT repeat the same insight in different wording.\n- Keep explanations short and concrete.\n- Maintain a neutral, analytical tone.\n",
"options": {},
"promptType": "define"
},
"typeVersion": 3
},
{
"id": "bf883fd6-ee8c-4bb5-ae8c-747c3bef9545",
"name": "Send Email (Report)",
"type": "n8n-nodes-base.gmail",
"position": [
9584,
-1248
],
"parameters": {
"sendTo": "user@example.com",
"message": "={{ $json.output }}",
"options": {
"appendAttribution": false
},
"subject": "=Weekly AI / LLM Intelligence Report \u2013 {{ new Date().toLocaleDateString() }}\n",
"emailType": "text",
"authentication": "serviceAccount"
},
"typeVersion": 2.1
},
{
"id": "466aa3a8-93ba-480a-abab-a151baecba40",
"name": "Search API (Decodo)",
"type": "@decodo/n8n-nodes-decodo.decodo",
"position": [
6832,
-1232
],
"parameters": {
"query": "={{ $json.search_query }}",
"operation": "google_search",
"results_limit": 1
},
"typeVersion": 1
},
{
"id": "161455d9-906d-4492-8a64-f9c4e3e7f29b",
"name": "Scrape Content (Decodo)",
"type": "@decodo/n8n-nodes-decodo.decodo",
"position": [
7376,
-1120
],
"parameters": {
"url": "={{$json.url}}"
},
"typeVersion": 1
},
{
"id": "9774fd35-6fe5-449e-adfc-c13a49fcbd82",
"name": "Set Search Config",
"type": "n8n-nodes-base.set",
"position": [
6448,
-1232
],
"parameters": {
"mode": "raw",
"options": {},
"jsonOutput": "{\n \"search_query\": \"latest AI LLM news\",\n \"max_results\": 10,\n \"min_relevance_score\": 50,\n \"summary_language\": \"en\",\n \"delivery_channel\": \"email\",\n \"debug_mode\": false\n}\n"
},
"typeVersion": 3.4
},
{
"id": "b92d812a-ee0a-4e69-98c1-2cdba7cfb437",
"name": "Check Content Quality",
"type": "n8n-nodes-base.if",
"position": [
7600,
-1104
],
"parameters": {
"options": {},
"conditions": {
"options": {
"version": 3,
"leftValue": "",
"caseSensitive": true,
"typeValidation": "strict"
},
"combinator": "and",
"conditions": [
{
"id": "606feef1-6b79-4a74-975d-25181e6c2af2",
"operator": {
"type": "string",
"operation": "notEmpty",
"singleValue": true
},
"leftValue": "={{$json.results?.[0]?.content}}",
"rightValue": ""
},
{
"id": "502031bd-4346-40c0-aee3-aefd744a99f9",
"operator": {
"type": "number",
"operation": "gt"
},
"leftValue": "={{$json.results?.[0]?.content.length}}",
"rightValue": 300
}
]
}
},
"typeVersion": 2.3
},
{
"id": "f267c3b2-6a23-4aed-a97c-bb91b8fdb56a",
"name": "OpenAI Model (Summary)",
"type": "@n8n/n8n-nodes-langchain.lmChatOpenAi",
"position": [
8208,
-1040
],
"parameters": {
"model": {
"__rl": true,
"mode": "list",
"value": "gpt-4.1-mini"
},
"options": {},
"builtInTools": {}
},
"typeVersion": 1.3
},
{
"id": "150b5b6b-cfc1-4cc9-8b2f-623f9cd4a81f",
"name": "Check Results Exist",
"type": "n8n-nodes-base.if",
"position": [
8752,
-1248
],
"parameters": {
"options": {},
"conditions": {
"options": {
"version": 3,
"leftValue": "",
"caseSensitive": true,
"typeValidation": "strict"
},
"combinator": "and",
"conditions": [
{
"id": "4944b6a5-9ddc-4116-a95c-d418f686546c",
"operator": {
"type": "number",
"operation": "gt"
},
"leftValue": "={{ $json.summaries.length }}",
"rightValue": 0
}
]
}
},
"typeVersion": 2.3
},
{
"id": "a44050d8-28b8-4da0-b907-9631058f8c3d",
"name": "Send Email (Empty)",
"type": "n8n-nodes-base.gmail",
"position": [
8864,
-1072
],
"parameters": {
"sendTo": "user@example.com",
"message": "=No Updates Found...",
"options": {
"appendAttribution": false
},
"subject": "=Weekly AI / LLM Intelligence Report \u2013 {{ new Date().toLocaleDateString() }}\n",
"emailType": "text",
"authentication": "serviceAccount"
},
"typeVersion": 2.1
},
{
"id": "11719afc-dcc6-4435-a212-574152fe9a2b",
"name": "OpenAI Model (Agent)",
"type": "@n8n/n8n-nodes-langchain.lmChatOpenAi",
"position": [
9136,
-1024
],
"parameters": {
"model": {
"__rl": true,
"mode": "list",
"value": "gpt-4.1-mini"
},
"options": {},
"builtInTools": {}
},
"typeVersion": 1.3
},
{
"id": "795a8aae-0023-4a07-8d64-4dcdc47cb821",
"name": "Telegram Error Alert",
"type": "n8n-nodes-base.telegram",
"position": [
7664,
-864
],
"parameters": {
"text": "Warning ... Error in scraping ",
"chatId": "YOUR_CHAT_ID",
"additionalFields": {}
},
"typeVersion": 1.2
}
],
"active": false,
"settings": {
"executionOrder": "v1"
},
"versionId": "8b7b7448-16b3-44ed-9a07-3ecd8679ccad",
"connections": {
"Batch Loop": {
"main": [
[
{
"node": "Summarize Chain",
"type": "main",
"index": 0
}
],
[
{
"node": "Scrape Content (Decodo)",
"type": "main",
"index": 0
}
]
]
},
"Sanitize HTML": {
"main": [
[
{
"node": "Batch Loop",
"type": "main",
"index": 0
}
]
]
},
"Weekly Trigger": {
"main": [
[
{
"node": "Set Search Config",
"type": "main",
"index": 0
}
]
]
},
"Summarize Chain": {
"main": [
[
{
"node": "Aggregate Summaries",
"type": "main",
"index": 0
}
]
]
},
"Set Search Config": {
"main": [
[
{
"node": "Search API (Decodo)",
"type": "main",
"index": 0
}
]
]
},
"Aggregate Summaries": {
"main": [
[
{
"node": "Check Results Exist",
"type": "main",
"index": 0
}
]
]
},
"Check Results Exist": {
"main": [
[
{
"node": "Research Analyst Agent",
"type": "main",
"index": 0
}
],
[
{
"node": "Send Email (Empty)",
"type": "main",
"index": 0
}
]
]
},
"Filter & Parse URLs": {
"main": [
[
{
"node": "Batch Loop",
"type": "main",
"index": 0
}
]
]
},
"Search API (Decodo)": {
"main": [
[
{
"node": "Filter & Parse URLs",
"type": "main",
"index": 0
}
]
]
},
"OpenAI Model (Agent)": {
"ai_languageModel": [
[
{
"node": "Research Analyst Agent",
"type": "ai_languageModel",
"index": 0
}
]
]
},
"Check Content Quality": {
"main": [
[
{
"node": "Sanitize HTML",
"type": "main",
"index": 0
}
],
[
{
"node": "Telegram Error Alert",
"type": "main",
"index": 0
}
]
]
},
"OpenAI Model (Summary)": {
"ai_languageModel": [
[
{
"node": "Summarize Chain",
"type": "ai_languageModel",
"index": 0
}
]
]
},
"Research Analyst Agent": {
"main": [
[
{
"node": "Send Email (Report)",
"type": "main",
"index": 0
}
]
]
},
"Scrape Content (Decodo)": {
"main": [
[
{
"node": "Check Content Quality",
"type": "main",
"index": 0
}
]
]
}
}
}
Pro
For the full experience including quality scoring and batch install features for each workflow upgrade to Pro
About this workflow
Staying up-to-date with AI and LLM developments requires reading dozens of articles every week. Manual research is time-consuming and often leads to “information overload” or reading low-quality clickbait. Important technical breakthroughs often get buried under marketing fluff.
Source: https://n8n.io/workflows/12074/ — original creator credit. Request a take-down →
Related workflows
Workflows that share integrations, category, or trigger type with this one. All free to copy and import.
This workflow automates the process of generating, reviewing, and publishing blog posts across multiple platforms, now enhanced with support for RSS Feeds as a content source. It streamlines the manag
HTTP Request, Html Extract, RSS Feed Read +9
This n8n workflow automates the daily generation of comprehensive analytics reports from multiple websites, processes them using OpenAI's powerful language models, and then delivers the insights direc
Memory Buffer Window, Google Analytics, Agent +6
This n8n workflow automatically generates AI-powered content about local news and publishes it across multiple social media platforms. The workflow runs on a schedule, fetches the latest news about a
Edit Image, WordPress, HTTP Request +10
Decodo is a powerful public data access platform offering managed web scraping APIs and proxy infrastructure to collect structured web data at scale. It handles proxies, anti-bot protection, JavaScrip
OpenAI Chat, Agent, Slack +5
//ASMR AI Workflow
HTTP Request, Tool Think, OpenAI Chat +6