Master 80 of n8n in 36 Minutes

The video provides a comprehensive guide to n8n, an automation platform, focusing on mastering its most frequently used features. It demonstrates building three automations: a simple daily weather email, a form-driven inquiry router, and an advanced AI personal assistant. The tutorial also covers n8n's core concepts, node types, self-hosting options, and the distinction between automations and dynamic agents.

Introduction to n8n and Core Concepts #

• What is n8n?
  • Node-based automation platform with growing popularity, especially for AI agents.
• Automations vs. Agents
  • Automations: Fixed, set sequence (A to B to C), even with branches or AI.
  • Agents: Dynamic, can reason, make decisions, and choose actions based on context.
• Use Cases for n8n
  • Personal Productivity: AI assistants for email, calendar, documents.
  • Marketing & Business: Automate social media, CRM, customer onboarding, sales processes.
  • E-commerce: Sync product data, process orders, handle customer support.
  • Data & Reporting: Pull, clean, and generate reports from multiple sources.
  • Home & Lifestyle: Smart home integration, time/location-based routines.
  • AI Integrations: Connect language models, image tools, custom APIs for summarization, content creation, complex agents.

Building Automation 1: Daily Weather Email #

• Objective: Create a simple automation to run daily, check the weather, and email it.
• Workflow Setup:
  • New workflow named "daily weather."
  • Trigger Node (Schedule):
    • Sets workflow to run daily at 6 AM.
    • Trigger interval: days, 1 day between, trigger hour: 6, minutes: 0.
  • Action Node (OpenWeather):
    • Connects to Open Weather API to get current weather data.
    • Credentials Setup:
      • Create an account on openweather.org.
      • Copy API key from "API keys" section.
      • Paste into n8n's access token field.
    • Settings:
      • Current weather, Imperial units, By zip code (enter zip code).
    • Testing: Use "Test Step" to verify data retrieval.
    • Data Structure: Left side (input from previous node), middle (settings), right side (output to next node). Output data viewable as schema, table, or JSON.
  • Action Node (Gmail):
    • Sends an email with weather information.
    • Credentials Setup: Sign in with Google, approve access.
    • Settings:
      • Recipient email (personal email).
      • Subject: "Daily weather."
      • Email type: Text.
      • Message populated dynamically by dragging data fields (temperature, location, min/max temp, wind speed) from the OpenWeather output.
      • Disable "append nad attribution" option.
  • Final Steps:
    • Save regularly.
    • "Test Workflow" to run end-to-end.
    • Check email for successful delivery.
• Canvas Tips:
  • Broom icon to tidy up layout.
  • Zoom in/out, "zoom to fit" to center.
  • Executions Tab: View past workflow runs, success/failure status.
  • "Copy to editor" to revert to previous version.
  • Activate workflow (toggle "active") to run automatically.

Account Setup & Self-Hosting n8n #

• n8n Cloud: Easiest option, $20/month after free trial.
• Self-Hosting: More control, lower cost, two methods:
  • Local Install with Docker: More technical, not covered in video.
  • Virtual Private Server (VPS): Easier middle ground, used by presenter.
  • Hostinger VPS (Sponsor):
    • One-click installation template for n8n.
    • Select "self-hosted NAD option" under VPS services.
    • KVM2 plan recommended.
    • Use coupon code "Futuredia."
    • Manage app, activate free license key (sent via email).

Core Node Types #

• Categories:
  • Triggers: Initiate workflows (schedule, form, chat input, webhook).
  • Actions: Pre-built integrations (Google products, weather, NASA, Notion, Airtable, Spotify, Stripe, etc.).
  • HTTP Request: Universal adapter for connecting to any API not on the pre-built list.
  • Logic Nodes: Control data flow (filter, branch, merge, loop – e.g., "if yes, do this; if no, do that").
  • AI Agent Node: Powers final build, connects to LLMs, adds memory, attaches tools.

Building Automation 2: Form-Driven Inquiry Router #

• Objective: Route sponsorship inquiries based on budget, auto-respond via email, and log to Google Sheet.
• Workflow Setup:
  • New workflow.
  • Trigger Node (Form Submission):
    • Uses n8n's default test forms (or any form platform like HubSpot).
    • Add fields: Company Name, Contact Name, Email, Website, Budget (number), Notes (text area).
    • Pin Data: After filling out and submitting the test form, pin the data to avoid re-entering for testing.
  • Logic Node (Switch):
    • Routes forms based on budget size.
    • Drag "budget" field into condition.
    • Rule 1: "is less than" 1 million (for "lowball" offers).
    • Rule 2: "is greater than or equal to" 1 million (for "serious" offers).
    • Creates two separate branches.
  • Action Node (Gmail - for both branches):
    • Credentials Setup: More steps for self-hosted Google credentials (documentation link provided). Client ID and Client Secret needed.
    • Settings:
      • Recipient: Drag "email" field from form input.
      • Subject: Dynamic (e.g., "Hello [company name]").
      • Message: Dynamic (e.g., "Sam Altman, how dare you waste my time..." or "I'm listening...").
      • Disable n8n attribution.
    • Duplicating Nodes: Duplicate the first Gmail node for the second branch and adjust message.
  • Action Node (Google Sheets - for both branches):
    • Add a new row with form data.
    • Credentials Setup: Add Google Sheets API from Google Console, enable, use same Client ID/Secret.
    • Sheet Creation: Create a new Google Sheet (e.g., "sponsor inquiry tracker") with columns for all form fields + "action."
    • Settings:
      • Document: Paste sheet URL.
      • Sheet: Select "sheet one."
      • Populate columns: Drag data fields from form input.
      • "Action" column: Manually enter "decline" or "follow up sent."
    • Test step.
    • Duplicate for the second branch and adjust the "action" text.
  • Refinements:
    • Rename nodes for clarity.
    • Rename switch node paths.
    • Unpin data and "clear execution" before full workflow test.
    • Test with different budget amounts to verify branching.
• Expansion Ideas:
  • Enrich Lead: Use AI agent to research company, summarize, or pull LinkedIn profiles/scrape sites.
  • Merge Nodes: Combine data from multiple sources (e.g., scraping multiple subreddits).
  • Set Nodes: Clean or reshape data (e.g., remove dollar signs from budget field if it was a text input).

Understanding JSON #

• JSON (JavaScript Object Notation): How data moves through nodes.
• $ JSON: Refers to data from the previous node (e.g., $json.budget means the "budget" field from the previous node's output).
• Specific Node Data: For data from earlier nodes, specifies the node name (e.g., {{$json["On Form Submission"].company}} means "company" field from "On Form Submission" node).
• Brackets: Used for fields with spaces (e.g., ["Company Name"]).
• Practical Use: Drag and drop fields are primary; no need to write JSON from scratch. ChatGPT can explain/generate snippets.
• Advanced Expressions: math.round() for cleaning numbers; more complex expressions for data cleaning (ask ChatGPT).

AI Agent Node, HTTP Requests & APIs #

• Agent Types:
  • Automation: Static, rule-based, fixed path.
  • Agent: Dynamic, flexible, reasons, plans, decides tools.
• Agent Core Components:
  • Brain: Large Language Model (LLM) for reasoning and language (ChatGPT, Claude, Gemini).
  • Memory: Provides context, remembers past interactions (conversation history, external docs, vector database).
  • Tools: How the agent interacts with the outside world (retrieve data, take action, orchestrate other workflows/agents).
• HTTP Requests & APIs:
  • API (Application Programming Interface): How software communicates (like vending machine buttons).
  • GET Request: Pulls information (e.g., weather, video, news feed).
  • POST Request: Sends information (e.g., form submission, add to sheet, prompt to LLM).
  • HTTP Request Node: Used for APIs not covered by built-in integrations.
  • Example (Air Quality Data):
    • airnow.gov has air quality API but no built-in n8n integration.
    • Use HTTP Request node (GET request).
    • Find API documentation, generate URL (e.g., query tool on airnow.gov).
    • Paste URL into n8n's HTTP Request node.
    • Combine with existing weather data using a merge node for a single email.

Building AI Personal Assistant Agent #

• Objective: Create an AI agent that manages calendar, emails, and documents, with custom tools for Reddit, YouTube, and video generation.
• Workflow Setup:
  • Trigger Node (Chat Trigger): For interacting within n8n.
    • Can be later swapped for Slack, WhatsApp, Telegram, Discord.
  • AI Agent Node:
    • Brain (LLM):
      • Select OpenAI.
      • Credentials: Go to platform.openai.com/s, add funds, create API key, copy immediately, paste into n8n.
      • Model: GPT4o.
    • Memory: Simple memory (e.g., 10 messages) for built-in chat.
    • Tools:
      • Google Calendar: Set read-only permission (summarize schedule) or full permission (add/change events).
      • Gmail: Set read-only (summarize emails) or full permission (send emails).
      • Google Sheet: Connect to the sponsor intake sheet (same selection as before).
    • System Prompt:
      • Click "Add Options," then "System Message."
      • Content (ask ChatGPT to generate): Role (personal assistant), instructions, tools available and when to use them, relevant context, restrictions.
• Testing Initial Agent:
  • Open chat in n8n.
  • Ask "What do you think of NAD?" (tests LLM).
  • Ask "Summarize my recent emails." (tests Gmail tool).
  • Ask "Do I have any good sponsorship requests?" (tests Google Sheet tool).
  • Ask "What openings do I have on my calendar tomorrow?" (tests Google Calendar tool).
  • Ask "Draft an email response for me and include the two options I have free for meetings during the day." (tests LLM + tools).
• Advanced Tool Integration (Calling Other Workflows/Agents):
  • Problem: Adding all tools into one agent makes it hard to organize and for agent to reason.
  • Solution: Create separate workflows (sub-agents) for specific tasks (e.g., calendar agent, Gmail agent) that the main agent can call.
  • Example Calendar Agent: Separate workflow with trigger "When executed by another workflow."
  • Integrating Sub-Agents:
    • Delete original Google Calendar tool from main agent.
    • Add "Call n8n Workflow" tool.
    • Choose calendar agent from list.
    • Input fields: Use {{$json["parameter_name"]}} or click stars for model to define automatically.
    • Update system prompt to inform agent about the new sub-agent.
• Full Personal Assistant Demo:
  • Main agent has calendar, Gmail, documents as sub-agents.
  • Additional custom tools: Reddit Curator, YouTube Summarizer, YouTube Shorts Creator (HTTP Request to another workflow).
  • All tools documented in the system prompt.
  • Demo Interactions:
    • "What are the comments like on my recent YouTube video?" (triggers YouTube summarizer -> processes with LLM).
    • "What are the best recent posts about AI on Reddit?" (triggers Reddit curator -> scrapes multiple subreddits -> processes with LLM).
    • "Post an AISMR short to my YouTube channel." (triggers YouTube Shorts creator workflow via webhook for async processing).

Final Tips #

• Expect Errors: Part of the process, test, iterate, adjust.
• Use ChatGPT as Build Partner: Train it on n8n docs.
• Keep Exploring:
  • Retrieval Augmented Generation (RAG): Uses vector databases (e.g., Pine Cone) for larger knowledge bases beyond chat history or Google Docs.
  • n8n Templates: Thousands available to learn from and build upon.
  • Futuredia Course Platform: Over 500 lessons on AI (ChatGPT, prompt engineering, custom GPTs, etc.).