by PiAPI
Who is the template for? This workflow is specifically designed for content creators and social media professionals, enabling Instagram and X (Twitter) influencers to produce highly artistic visual posts, empowering marketing teams to quickly generate event promotional graphics, assisting blog authors in creating featured images and illustrations, and helping knowledge-based creators transform key insights into easily shareable card visuals. Set up Instructions Fill in your API key from PiAPI. Fill in Basic Params Node following the sticky note guidelines. Set up a design template in Canvas Switchboard. Make a simple template in Switchboard. Click Crul and get the API code to fill in JSON of Design in Canvas. Click Test Workflow and get a url result. Use Case Here we will provide some setting examples to help users find a proper way to use this workflow. User could change these settings based on specific purposes. Basic Params Setting: theme: Hope scenario: Don't know about the future, confused and feel lost with tech-development. style: Cinematic Grandeur, Sci-Tech Aesthetic, 3D style example: 1. March. Because of your faith, it will happen. 2. Something in me will save me. 3. To everyone carrying a heavy heart in silence. You are going to be okay. 4. Tomorrow will be better. image prompt: A cinematic sci-fi metropolis where Deep Neural Nets control a hyper-connected society. Holographic interfaces glow in the air as robotic agents move among humans, symbolizing Industry 4.0. The scene contrasts organic human emotion with cold machine precision, rendered in a hyper-realistic 3D style with futuristic lighting. Epic wide shots showcase the grandeur of this civilization’s industrial evolution. Output Image: More Example Results for Reference
by Omer Fayyaz
An intelligent AI-powered agent that automatically browses publication websites, analyzes page content with natural language understanding, and identifies the latest downloadable reports, research papers, and data files across multiple sources using advanced structured output parsing. What Makes This Different: AI-Powered Content Analysis** - Uses advanced language models (GPT-4/GPT-5.1) to understand page context and identify downloadable reports, even when links aren't explicitly labeled, handling complex page layouts and dynamic content Structured Output Parsing** - Enforces JSON schema validation ensuring consistent data extraction with required fields (title, link, file_type, description), eliminating parsing errors and data inconsistencies HTML to Markdown Conversion** - Converts raw HTML to clean Markdown before AI processing, removing noise and improving AI comprehension of page structure and content hierarchy Intelligent Link Detection** - AI agent identifies direct download URLs, converts relative links to absolute URLs, and prioritizes the most recent reports based on publication dates and page positioning Comprehensive Validation** - Multi-layer validation checks link format, file type detection, and report relevance before saving, ensuring only valid, downloadable reports enter your library Flexible Source Management** - Reads publication sources from Google Sheets, enabling easy addition/removal of sources without workflow modification, with support for categories and custom metadata Key Benefits of AI-Powered Report Discovery: Automated Discovery** - Eliminates manual browsing and searching across multiple publication sites, saving hours of research time while ensuring you never miss new reports Context-Aware Extraction** - AI understands page context, distinguishing between actual reports and navigation links, category pages, or promotional content Prioritized Results** - Automatically selects the most recent and relevant report from each source, focusing on quality over quantity Structured Data Output** - All discovered reports are saved with consistent metadata (title, link, file type, description, source), making them easy to search, filter, and integrate with other systems Error Resilience** - Handles missing reports gracefully, logging when no reports are found without failing the entire workflow, ensuring continuous operation Integration Ready** - Can be called by other workflows (e.g., PDF downloader), enabling end-to-end automation from discovery to storage Who's it for This template is designed for researchers, market analysts, competitive intelligence teams, academic institutions, industry monitoring services, and anyone who needs to systematically discover and track downloadable reports from multiple publication sources. It's perfect for organizations that need to monitor industry publications, track competitor research, discover new market reports, build research libraries, or stay updated on latest publications without manually visiting dozens of websites daily. How it works / What it does This workflow creates an AI-powered report discovery system that reads publication source URLs from Google Sheets, fetches their pages, uses AI to analyze content, and extracts information about downloadable reports. The system: Reads Active Sources - Fetches publication URLs and metadata from Google Sheets "Report Sources" sheet, processing each source in sequence Loops Through Sources - Processes sources one at a time using Split in Batches, ensuring proper error isolation and preventing batch failures Fetches Publication Pages - Downloads HTML content from each source URL with proper browser headers (User-Agent, Accept, Accept-Language) to avoid blocking Converts HTML to Markdown - Transforms raw HTML into clean Markdown format, removing styling, scripts, and navigation elements to improve AI comprehension AI Analysis - LangChain agent analyzes the Markdown content using GPT-4/GPT-5.1, identifying downloadable reports based on context, link patterns, and content structure Structured Output Parsing - Enforces JSON schema validation, ensuring the AI returns data in the exact format: source, title, link, file_type, description Validates & Normalizes Output - Validates extracted links are absolute URLs, checks file type indicators, determines report validity, and normalizes all fields Routes by Validity - IF node routes valid reports to save operation, invalid/missing reports to logging Saves Discovered Reports - Appends valid reports to Google Sheets "Discovered Reports" sheet with metadata, source URL, category, and discovery timestamp Logs No Report Found - Records sources where no valid reports were found in "Discovery Log" sheet for monitoring and troubleshooting Tracks Completion - Generates completion summary with number of sources checked and processing timestamp Key Innovation: AI-Powered Context Understanding - Unlike traditional web scrapers that rely on fixed CSS selectors or regex patterns, this workflow uses AI to understand page context and semantics. The AI can identify reports even when they're embedded in complex layouts, use non-standard naming, or require understanding of surrounding text to determine relevance. This makes it adaptable to any website structure without manual configuration. How to set up 1. Prepare Google Sheets Create a Google Sheet with three tabs: "Report Sources", "Discovered Reports", and "Discovery Log" In "Report Sources" sheet, create columns: Source_Name, Source_URL, Category (optional) Add publication URLs in the Source_URL column (e.g., "https://example.com/research" or "https://publisher.com/reports") Add descriptive names in Source_Name column for easy identification Optionally add Category values (e.g., "Market Research", "Industry Reports", "Academic Papers") The "Discovered Reports" sheet will be automatically populated with columns: source, title, link, fileType, description, sourceUrl, category, discoveredAt, status, isValid The "Discovery Log" sheet will record sources where no reports were found Verify your Google Sheets credentials are set up in n8n (OAuth2 recommended) 2. Configure Google Sheets Nodes Open the "Read Active Sources" node and select your spreadsheet from the document dropdown Set sheet name to "Report Sources" Configure the "Save Discovered Report" node: select same spreadsheet, set sheet name to "Discovered Reports", operation should be "Append or Update" Configure the "Log No Report Found" node: same spreadsheet, "Discovery Log" sheet, operation "Append or Update" Test connection by running the "Read Active Sources" node manually to verify it can access your sheet 3. Set Up OpenAI Credentials Open the "OpenAI GPT-5.1" node (or configure the model you want to use) Connect your OpenAI API credentials (API key required) The workflow uses GPT-5.1 by default, but you can change to GPT-4, GPT-4 Turbo, or other models Temperature is set to 0.1 for consistent, deterministic output Verify API key has sufficient credits and access to the selected model For cost optimization, GPT-4 Turbo is recommended for similar results at lower cost 4. Configure AI Agent & Output Parser The "AI Report Discovery Agent" node contains a detailed system prompt that instructs the AI on what to look for The prompt is pre-configured but can be customized for your specific needs (e.g., prioritize certain file types, look for specific keywords) The "Structured Output Parser" enforces the JSON schema - verify the schema matches your needs: { "source": "Publisher Name", "title": "Report Title", "link": "https://example.com/report.pdf", "file_type": "pdf", "description": "Brief description" } The parser ensures the AI always returns valid JSON with all required fields Test the AI agent by manually running with a sample source URL to verify it correctly identifies reports 5. Customize Discovery Rules (Optional) The AI agent's system prompt can be modified in the "AI Report Discovery Agent" node Current rules prioritize: downloadable files (PDF, Excel, Word, PowerPoint), most recent publications, direct download URLs To customize: Edit the system message to add specific keywords, file types, or discovery patterns Example customization: Add industry-specific terms or prioritize reports with certain keywords in titles The validation code in "Validate & Normalize Output" can be adjusted to change what's considered "valid" Test with your specific sources to ensure discovery rules work as expected 6. Set Up Scheduling & Test The workflow includes Manual Trigger (for testing), Schedule Trigger (runs daily), and Execute Workflow Trigger (for calling from other workflows) To customize schedule: Open "Schedule (Daily)" node and adjust interval (e.g., twice daily, weekly) For initial testing: Use Manual Trigger, add 2-3 test publication URLs to your "Report Sources" sheet Verify execution: Check that pages are fetched, AI analysis completes, and reports are saved to "Discovered Reports" Monitor execution logs: Check for API errors, timeout issues, or parsing failures Review Discovery Log: Verify sources with no reports are properly logged Common issues: OpenAI API rate limits (add delays if processing many sources), invalid URLs (check source URLs), timeout errors (increase timeout for slow-loading pages), AI not finding reports (may need to adjust system prompt for specific site structures) Requirements OpenAI API Key** - Active OpenAI account with API access and sufficient credits for GPT-4/GPT-5.1 model usage (API key configured in n8n credentials) Google Sheets Account** - Active Google account with OAuth2 credentials configured in n8n for reading and writing spreadsheet data Source Spreadsheet** - Google Sheet with "Report Sources", "Discovered Reports", and "Discovery Log" tabs, properly formatted with required columns Valid Publication URLs** - Direct links to publication pages that contain downloadable reports (not direct PDF links - the workflow discovers those) n8n Instance** - Self-hosted or cloud n8n instance with access to external websites (HTTP Request node needs internet connectivity) and LangChain nodes enabled
by Lucas Walter
Who's it for This template is perfect for sales professionals, marketers, and business developers who need to quickly gather contact information from company websites. Whether you're building prospect lists, researching potential partners, or collecting leads for outreach campaigns, this automation saves hours of manual email hunting. What it does This workflow automatically discovers and extracts email addresses from any website by: Taking a website URL as input through a simple form Using Firecrawl's mapping API to find relevant pages (about, contact, team pages) Batch scraping those pages to extract email addresses Intelligently handling common email obfuscations like "(at)" and "(dot)" Returning a clean, deduplicated list of valid email addresses The automation handles rate limiting, retries failed requests, and filters out invalid or hidden email addresses to ensure you get quality results. How to set up Get Firecrawl API access: Sign up at firecrawl.dev and obtain your API key Configure credentials: In n8n, create a new HTTP Header Auth credential named "Firecrawl" with: Header Name: Authorization Header Value: Bearer YOUR_API_KEY Import the workflow: Copy the workflow JSON into your n8n instance Test the form: Activate the workflow and test with a sample website URL How to customize the workflow Search parameters: Modify the search parameter in the map_website node to target different page types (currently searches for "about contact company authors team") Extraction limits: Adjust the limit parameter to scrape more or fewer pages per website Retry logic: The workflow includes retry logic with a 12-attempt limit - modify the check_retry_count node to change this Output format: The set_result node formats the final output - customize this to match your preferred data structure Email validation: The JSON schema in start_batch_scrape defines how emails are extracted - modify the prompt or schema for different extraction rules The workflow is designed to be reliable and handle common edge cases like rate limiting and failed requests, making it production-ready for regular use.
by Yaron Been
Wan Video Wan 2.2 I2v A14b Video Generator Description Image-to-video at 720p and 480p with Wan 2.2 A14B Overview This n8n workflow integrates with the Replicate API to use the wan-video/wan-2.2-i2v-a14b model. This powerful AI model can generate high-quality video content based on your inputs. Features Easy integration with Replicate API Automated status checking and result retrieval Support for all model parameters Error handling and retry logic Clean output formatting Parameters Required Parameters prompt** (string): Prompt for video generation image** (string): Input image to generate video from Optional Parameters seed** (integer, default: None): Random seed. Leave blank for random num_frames** (integer, default: 81): Number of video frames. 81 frames give the best results resolution** (string, default: 480p): Resolution of video. 832x480px corresponds to 16:9 aspect ratio, and 480x832px is 9:16 sample_shift** (number, default: 5): Sample shift factor sample_steps** (integer, default: 30): Number of generation steps. Fewer steps means faster generation, at the expensive of output quality. 30 steps is sufficient for most prompts frames_per_second** (integer, default: 16): Frames per second. Note that the pricing of this model is based on the video duration at 16 fps How to Use Set up your Replicate API key in the workflow Configure the required parameters for your use case Run the workflow to generate video content Access the generated output from the final node API Reference Model: wan-video/wan-2.2-i2v-a14b API Endpoint: https://api.replicate.com/v1/predictions Requirements Replicate API key n8n instance Basic understanding of video generation parameters
by Elliot Scribner
> Disclaimer: this workflow template uses the n8n-nodes-couchbase community package. Community nodes are unverified and usage of them comes with some risks. See here for instructions on installing n8n community nodes. This template is intended for use by those interested in learning more about Agentic AI workflow development, as well as those interested in learning how to use the Couchbase Search Vector Store node for practical applications. This workflow helps users decide on travel destinations based on descriptions of several points of interest loaded into Couchbase and retrieved using Vector Search. How it Works This template contains two workflows: The Data Ingestion workflow uses the following nodes Webhook node (to listen for HTTP requests) OpenAI Embeddings node (to generate embeddings on document insertion) Note: You’ll need to configure OpenAI credentials for this node Couchbase Vector node (configured for document insertion) Default Data Loader and Recursive Character Text Splitter The Chat Application workflow uses the following nodes Chat Trigger node AI Tools Agent node connect to: Gemini (as the Chat Model, for generating responses) Note: You will have to configure Gemini credentials for this node Simple Memory (as the Memory, to maintain conversation context) Couchbase Search Vector node (as the Tool, for search) OpenAI Embeddings node (as the Embedding model for the Couchbase Search Vector node, to convert queries to vectors) Note: You’ll need to configure OpenAI credentials for this node Set up Setting up this workflow is easy and only takes around 10 minutes. Prerequisites A Couchbase Cluster running the Search Service, and corresponding database access credentials Be sure the Couchbase cluster allows the incoming IP address for n8n Create a Vector Search Index using this index definition Create a bucket (called travel-agent), scope (called vectors), and collection (called points-of-interest) in your Cluster OpenAI API Key Gemini API Key Steps Configure all necessary credentials (Couchbase, OpenAI, and Gemini) Select your bucket, scope, and collection for each of the Couchbase vector nodes Ingest data, either using the cURL statements found on the sticky note within the workflow, or using this shell script to ingest 6 points of interest Open the chat and test out your travel agent! Customization and Next Steps This workflow template can be made more robust by enhancing the data model to include more information about each point of interest. For example, the addition of price ranges, ideal seasons to visit, activity types, and accomodation options can help inform the LLM further about each destination, and in turn allow it to provide a more tailored response and be more helpful for travel planning. Alternatively, the data model could be entirely re-configured to suit a wide variety of other use cases. This template can serve as a building block for all sorts of AI Agent applications using RAG and is not limited to only travel recommendations.
by Davide
This workflow implements an AI-powered WhatsApp booking assistant for a hair salon. The system allows customers to book, reschedule, or cancel appointments automatically via text or voice messages on WhatsApp. The workflow supports both text and voice messages. An AI agent manages the conversation logic, determines the user's intent, and calls the appropriate tools to perform actions such as checking availability, creating appointments, rescheduling bookings, or canceling events. The system also includes guardrails to filter unsafe or irrelevant inputs and escalation mechanisms to notify the team when a request falls outside supported tasks. Key Advantages 1. ✅ 24/7 Automated Appointment Booking Customers can book, reschedule, or cancel appointments at any time through WhatsApp without requiring manual staff intervention. 2. ✅ Multi-Modal Communication (Text + Voice) The workflow supports both text messages and voice messages. Audio messages are automatically transcribed and processed by the AI assistant. 3. ✅ Intelligent AI Assistant An AI agent manages the conversation, understands user intent, and automatically performs actions such as checking calendar availability or creating appointments. 4. ✅ Real-Time Calendar Integration The assistant connects directly to Google Calendar to: check available time slots create new bookings update existing appointments cancel reservations This ensures real-time accuracy. 5. ✅ Automatic Client Management Customer information such as phone number, name, and service type is stored and updated in Google Sheets, creating a lightweight CRM database. 6. ✅ Conversation Memory The system stores chat history in PostgreSQL, allowing the AI assistant to maintain context during conversations and deliver a more natural experience. 7. ✅ Smart Error Handling and Escalation If the AI cannot handle a request (for example, a question unrelated to appointments), the workflow automatically: notifies the team via email escalates the conversation for human assistance. 8. ✅ Secure and Controlled AI Responses Guardrails are implemented to prevent irrelevant or unsafe responses, ensuring that the assistant follows defined policies and tasks. 9. ✅ Omnichannel Architecture The workflow is designed to work with: WhatsApp messages chatbot interface (chat trigger) This allows the same logic to power multiple communication channels. 10. ✅ Scalable Automation The system is fully automated and can handle multiple simultaneous conversations, reducing operational workload and improving customer service efficiency. How it works Input Handling (WhatsApp & Chat): It listens for incoming messages from either a WhatsApp Trigger or a Chat Trigger. WhatsApp: It processes both text messages and voice notes. If a voice note is received, it automatically downloads the audio, transcribes it using OpenAI (Whisper), and converts it into text for the AI to process. The user's phone number (for WhatsApp) or session ID (for Chat) is stored as a sessionId to maintain conversation context. Guardrails (Security & Policy): Before the message reaches the main AI agent, it passes through a Guardrails node (using Google Gemini). This acts as a safety filter to check for jailbreak attempts or inappropriate content, ensuring the conversation stays within company policies. AI Agent (The Virtual Assistant): The cleaned-up message is sent to the AI Agent ("Virtual Assistant"). The agent is prompted to act as "Emma" and follows strict instructions: Identify User: It first uses the New client tool to check a Google Sheet to see if the user is already registered. Manage Appointments: Based on the user's request (book, reschedule, cancel), it intelligently calls specific tools in the correct order. Date Logic: It is aware of the current date and time to calculate correct future dates (e.g., interpreting "this Thursday"). Tool Execution (Actions): The agent has access to several tools to perform actions: Google Sheets (New client, Add client): To check if a client exists and add new clients to a spreadsheet. Google Calendar (Get events, Create event, Update event, Delete event): To check availability and manage appointments. The Get events tool is configured to look for slots in the next 30 days. Gmail (Send Email): To notify the team if a user asks for something outside the scope (e.g., pricing, product questions). WhatsApp HITL / Chat HITL (Escalation): To hand over the conversation to a human agent for complex requests. Response Delivery: The AI agent's text response is routed back to the user. On WhatsApp: It intelligently handles the response format. If the original message was text, it sends a text reply. If the original message was a voice note, it uses OpenAI's TTS to convert the text reply into an audio message and sends that back. Setup steps To get this workflow running, you need to configure the following credentials and nodes: Configure Credentials: Google Calendar OAuth2: Connect your Google account (used in Calendar nodes). Google Sheets OAuth2: Connect your Google account to access the spreadsheet. Gmail OAuth2: Connect your Google account to send notification emails from xxx@xxx.com. WhatsApp API: Connect your WhatsApp Business API credentials (Phone Number ID). OpenAI API (OpenRouter): Provide your API key for audio transcription and text-to-speech. Google Gemini API: Provide your API keys for the main AI model (Google Gemini Chat Model) and the Guardrails model (Google Gemini Chat Model1). Postgres: Configure the connection to your Postgres database for chat memory. Configure Google Sheets: Create a Google Sheet. Update the Document ID in the New client and Add client nodes with your sheet's ID. Ensure the sheet has columns for Phone Number, Client, and Service. Configure Google Calendar: Verify that the calendar ID in the Calendar nodes is correct and accessible. Configure WhatsApp Triggers & Actions: The WhatsApp Trigger node needs to be connected to your WhatsApp Business App. The Send message and Send audio nodes must have the correct Phone Number ID for your WhatsApp Business number. Update Email Recipient: In the Send Email tool (for escalations), change the sendTo email address from xxx@xxx.com to the actual email of the team that should receive these notifications. Review AI Prompts: Open the "Virtual Assistant" (AI Agent) node and review the system message. Update the salon name ("Cuts & Styles"), stylist name ("Alex Carter"), and address to match your business. Check the "Guardrails" node to adjust the safety threshold if necessary. 👉 Subscribe to my new YouTube channel. Here I’ll share videos and Shorts with practical tutorials and FREE templates for n8n. Need help customizing? Contact me for consulting and support or add me on Linkedin.
by Viktor Klepikovskyi
Advanced Retry and Delay Logic This template provides a robust solution for handling API rate limits and temporary service outages in n8n workflows. It overcomes the limitations of the default node retry settings, which cap retries at 5 and delays at 5 seconds. By using a custom loop with a Set, If, and Wait node, this workflow gives you complete control over the number of retries and the delay between them. Instructions: Replace the placeholder HTTP Request node with your target node (the one that might fail). In the initial Set Fields node, modify the max_tries value to set the total number of attempts for your workflow. Adjust the delay_seconds value to define the initial delay between retries. Optionally, configure the Edit Fields node to implement exponential backoff by adjusting the delay_seconds expression (e.g., {{$json.delay_seconds * 2}}). For a more detailed breakdown and tutorial of this template, you can find additional information here.
by Robert Breen
This guide walks you through building an intelligent AI Agent in n8n that routes tasks to the appropriate sub-agent using the new @n8n/n8n-nodes-langchain agent framework. You’ll create a Manager Agent that evaluates user input and delegates it to either an Email Agent or a Data Agent—each with its own role, memory, and OpenAI model. This is perfect for use cases where you want a single entry point but intelligent branching behind the scenes. 🔧 Step 1: Set Up the Manager Agent Start by dragging in an Agent node and name it something like ManagerAgent. This agent will act as the “brain” of your system, analyzing the user's input and determining whether it should be handled by the email-writing sub-agent or the data-summary sub-agent. Open the node’s settings and paste the following into the System Message: You are an AI Manager that delegates tasks to specialized agents. Your job is to analyze the user's message and decide whether it requires: An EmailAgent for writing outreach, follow-up, or templated emails, or A DataAgent for tasks involving data summaries, metrics, or analysis. Send the instructions to the sub agents. This instruction gives the Manager Agent clarity on what roles exist and what types of tasks belong to each one. 🧠 Step 2: Add Memory to the Manager Agent Drag in a Memory (BufferWindow) node and label it Manager Memory. Connect it to the ai_memory input of the Manager Agent. This ensures the agent can remember recent inputs and outputs from the user and agents during the conversation. No extra configuration is needed in this memory node—just connect it to the agent. 🔌 Step 3: Connect a Language Model to the Manager Agent Next, add a Language Model node and choose OpenAI Chat Model. Select a model like gpt-4o-mini or gpt-4, depending on what you have access to. Under Credentials, connect your OpenAI API key. If you haven’t created this credential yet: Click "OpenAI API" under Credentials. Choose "Create New". Paste your OpenAI API key (found at https://platform.openai.com/account/api-keys). Save it and return to the workflow. Once the model is set, connect it to the ai_languageModel input of the Manager Agent. ✉️ Step 4: Create the Email Agent Tool Now you’ll create a specialized sub-agent that only writes emails. Add an Agent Tool node and call it EmailAgent. In the tool’s settings, describe its job clearly. For example: Writes professional, friendly, or action-oriented emails based on instructions. Then scroll down to the System Message section and enter the following: You are a professional Email Writing Assistant. You write polished, effective emails for tasks such as outreach, follow-ups, and client communication. Follow the instruction provided exactly and return only the email content. Use a warm, business-appropriate tone. For the text input field, use the expression: {{ $fromAI('Prompt__User_Message_', ``, 'string') }} This allows the Email Agent to receive exactly what the Manager Agent wants it to handle. Add another Memory node and link it to this tool to help it maintain short-term context. Then add a second Language Model node, configured just like the first one (you can even clone it), and connect it to the EmailAgent. Finally, connect this entire EmailAgent setup back to the ManagerAgent by attaching it to its ai_tool input. 📊 Step 5: Create the Data Agent Tool Repeat the same steps, but this time for data summaries and analysis. Add another Agent Tool node and name it DataAgent. In the Tool Description, write something like: Responds to instructions requiring metrics, summaries, or data analysis explanations. For its input text field, you can use: {{json.query}} If desired, provide a system message that gives the agent more detailed instruction on how to behave: You are a helpful Data Analyst. Summarize trends, explain metrics, and break down data clearly based on user instructions. As with the EmailAgent, you’ll also need: A dedicated Memory node A dedicated Language Model node A connection to the ai_tool input of the Manager Agent Now the Manager Agent has two tools it can delegate to: one for communication and one for insights. 🧪 Step 6: Test Your AI Agent System Deploy the workflow and start testing by sending prompts like: > “Write a cold outreach email to a software company.” The ManagerAgent should route that to the EmailAgent. Then try: > “Summarize how our lead volume changed last month.” The DataAgent should receive that task. If routing isn’t working as expected, double-check your system messages and input bindings in each agent tool. ✅ You’re Done! You now have a modular, multi-agent AI system powered by n8n. The Manager Agent delegates intelligently, each sub-agent is optimized for its role, and all of them benefit from context memory. For more advanced setups, you can chain tools, add additional memory types, or use retrieval (RAG) tools for external document support.
by Khaled
🧾 Description: This automation uses GPT-4o to scan unread Gmail emails and intelligently classify them as: Action → Requires your attention (reply, review, schedule, or respond) No Action → Informational or promotional; no action needed The result? You eliminate inbox noise and gain a clear daily routine: only check what's in Action Required. ⚙️ How It Works: Trigger: Runs on a customizable schedule Fetch Emails: Pulls unread messages from Gmail Classify via GPT-4o: Determines if each email needs action or not Sort Emails: Labels actionable emails as Action Required Labels non-actionable ones as No Action Removes the Inbox label to clean your primary inbox view ✅ Emails stay in your account—just better organized 🚀 How to Use: Import the workflow into your n8n instance Set up Gmail and OpenAI credentials Create Gmail labels: Action Required No Action Activate the workflow Start your day by checking only the Action Required label 📦 Requirements: n8n (self-hosted or cloud) Gmail OAuth2 account OpenAI API key (GPT-4o or GPT-4o-mini) Gmail labels: Action Required, No Action 💡 Why It Matters: Stop manually filtering emails. This workflow helps you focus only on what matters while keeping everything else out of your way—without deleting or archiving anything.
by Niko
Capture URL Screenshots Automatically from Google Sheets & Drive with ScreenshotOne & Gmail Alerts Summary This automation template streamlines the process of capturing screenshots for multiple URLs. Instead of manually visiting each URL, taking a screenshot, and organizing the results, this workflow automates everything. When a spreadsheet is added to a designated Google Drive folder, the template extracts URLs from the column named "Url." These URLs are then processed through ScreenshotOne to capture screenshots, which are saved back to the same folder. Finally, an email notification is sent via Gmail with a link to the folder containing the screenshots. Problem Solved This template addresses the challenge of manual screenshot capture for multiple URLs. Without this automation, a user would need to: Open each URL from a spreadsheet. Take a screenshot manually. Save each screenshot with an appropriate name. Organize the screenshots in a folder. Notify stakeholders when the process is complete. These steps are not only time-consuming but also repetitive, especially when handling a large number of URLs. Who Can Benefit: Digital Marketers:** Monitor website appearances for competitive analysis or to track campaign landing pages. Web Developers/Designers:** Capture screenshots of multiple websites for inspiration or reference. QA Teams:** Document the visual state of web pages during various stages of development. SEO Specialists:** Track visual changes to websites they are optimizing. Content Managers:** Monitor how content appears across various web properties. Prerequisites Google Drive Node:** Must have appropriate permissions to create and access folders. Connected Google Sheets Node:** To extract URLs from the spreadsheet. Authenticated Gmail Node:** For sending notifications. ScreenshotOne Account:* Either a free or paid plan depending on volume needs, along with an *Access key**. Ensure you replace the placeholder --YOUR ACCESS KEY-- with your generated access key in the "Get Screenshots" node. Workflow Details Step 1: Google Drive Integration Trigger Node:** Monitors a specific folder in Google Drive. When a spreadsheet is added, the workflow is initiated. Step 2: Google Sheets Processing Google Sheets Node:** Extracts URLs from the column named "Url." Step 3: Screenshot Capture Get Screenshots Node:** Sends each extracted URL to ScreenshotOne to capture screenshots. Step 4: Saving Screenshots and Notifications Google Drive Node:** Saves the captured screenshots back into the same folder. Gmail Node:** Sends an email notification with a link to the folder, alerting stakeholders that the screenshots are ready. Customization Guidance Folder Monitoring: The workflow is set to monitor a specific Google Drive folder. It can be customized by selecting a different folder in the node settings. Spreadsheet Structure: While the template expects a spreadsheet with a column named "Url." for extracting URLs, users can add additional columns (e.g., titles, categories, or tags) and modify the workflow to utilize them as needed. Email Settings: Customize the recipient, subject, and body of the notification email to suit your needs. If required, enable optional notifications for different stakeholders. ScreenshotOne Access Key & Configurations: A valid ScreenshotOne Access key is required to capture screenshots. Users can further refine screenshot settings (e.g., viewport size, device emulation, or delay timing) by exploring the available options in the ScreenshotOne API documentation.
by Afnan
This n8n workflow automates the process of finding, summarizing, and posting breaking news headlines on X (formerly Twitter). It combines Google Custom Search for finding the latest news articles with Groq's LLaMA 3 model to generate short, engaging headlines — complete with hashtags — and posts them on your X account. 🔧 Features Custom topic support (e.g., "AI", "health", "technology") Automated scheduling every few hours Google Custom Search to find the most recent news articles Groq LLaMA3-based headline generation with hashtags Auto-post to X (Twitter) Built-in credential separation for API keys and access tokens 📦 Included Nodes Schedule Trigger Set (Set Topic, Google API Key, Custom Search CX, etc.) HTTP Request (Google Search API) Code Node (Format prompt and extract article data) HTTP Request (Groq API for headline generation) Twitter Node (Post to X) ⚙️ How It Works (Step-by-Step) Trigger The workflow starts on a scheduled interval (default: every 5 hours, at a random minute within the hour). Set Topic You can define your own topic keyword (e.g., AI, mental health, climate change) by editing the Set Topic node. Build Search Query Constructs a Google search query like: latest {topic} news. Google API Config Injects your own Google API Key and Custom Search CX (replace the placeholders in the Google Config node). Search for News Performs a real-time search using Google Custom Search API and fetches the latest article result. Generate Prompt for AI A JavaScript Function node extracts the top article’s title and link, formats it into a clean prompt including instructions to append hashtags. Groq AI Request Sends the prompt to Groq’s LLaMA 3 model to generate a concise, tweet-length headline with 1–2 relevant hashtags. Post to Twitter (X) The generated headline is posted to your connected X account via the Twitter OAuth2 API. ✅ Requirements Google API Key Google Custom Search Engine (CX) Groq API Key Twitter Developer App with OAuth2 credentials 💡 Customization Tips Change the topic in the Set Topic node to anything you like. Adjust the posting frequency in the Schedule Trigger node. Modify prompt behavior in the Function node to fit a specific tone or brand voice. Add logging, filtering, or multiple post variations as needed.
by Yaron Been
Zsxkib Canary Qwen 2.5b Text Generator Description 🎤The best open-source speech-to-text model as of Jul 2025, transcribing audio with record 5.63% WER and enabling AI tasks like summarization directly from speech✨ Overview This n8n workflow integrates with the Replicate API to use the zsxkib/canary-qwen-2.5b model. This powerful AI model can generate high-quality text content based on your inputs. Features Easy integration with Replicate API Automated status checking and result retrieval Support for all model parameters Error handling and retry logic Clean output formatting Parameters Required Parameters audio** (string): Audio file to transcribe Optional Parameters llm_prompt** (string, default: None): Optional LLM analysis prompt show_confidence** (boolean, default: False): Show AI reasoning in analysis include_timestamps** (boolean, default: True): Include timestamps in transcript How to Use Set up your Replicate API key in the workflow Configure the required parameters for your use case Run the workflow to generate text content Access the generated output from the final node API Reference Model: zsxkib/canary-qwen-2.5b API Endpoint: https://api.replicate.com/v1/predictions Requirements Replicate API key n8n instance Basic understanding of text generation parameters