How to Use n8n: From Your First Workflow to AI Agents

Are you still copying every new form response into an AI chat, asking for a summary, then pasting that summary into an email and a Slack message? It's fine for a few entries a day, but once responses pile up, all that copy-pasting quietly eats your whole morning.

n8n is an open-source automation tool that lets you lay out each step of a task as a node (a building block) and connect them with lines, like "when a form is submitted → AI summarizes it → send an email." You can build it visually without code, and you can even host it on your own server. Based on n8n's official docs as of October 2026, this guide covers how n8n differs from Zapier and Make, how to go from sign-up to your first workflow, how to connect a model and tools to the AI Agent node, and what to watch for with costs and personal data. At the end, we'll show how meeting notes organized with daglo can become the starting point for your automations.

What is n8n, and how is it different from Zapier and Make?

n8n is a tool for automating workflows by connecting blocks. The biggest differences from Zapier and Make are that you can self-host it and that it charges per workflow execution, not per step. The more steps your automation has, the more that pricing model works in your favor.

Pricing changes often, so check each service's pricing page before you sign up. As of October 2026, n8n's cloud Starter plan includes 2,500 executions per month, and every plan comes with unlimited users and workflows.

Getting started: cloud sign-up and the basic building blocks

If you're new, n8n Cloud is the fastest way in because there's nothing to install. Sign up at n8n.io with your email and a 14-day trial starts without a credit card, dropping you straight into the workflow editor.

• Workflow: The canvas that holds one automation. Think one task, one workflow.
• Trigger node: The condition that starts the automation, such as a form submission, a schedule, or a new email.
• Node: A block that does the actual work, like summarizing with AI, sending an email, or adding a row to a sheet.
• Credentials: The login details for outside services like Gmail, Slack, or OpenAI. Set them up once and reuse them across workflows.
• Executions: A log of when each workflow ran and what data passed through every step. This is the first place to look when something breaks.

If your company can't keep data in an outside cloud, look at the self-hosted Community Edition with Docker. You'll be responsible for server upkeep and security updates, though, so it's easier to learn the ropes in the cloud first and move later.

Build your first workflow: form response → AI summary → email and Slack alert

A great first automation is: when someone submits an inquiry form, AI summarizes it in three lines and alerts the right person by email and Slack. It takes one trigger, one AI node, and one or two notification nodes, so you learn every core piece of n8n.

1. Add a form trigger: In a new workflow, choose the n8n Form Trigger (On form submission) as the first step and add fields for name, contact, and message. n8n generates the form URL for you, so you don't need a separate survey tool.
2. Connect an AI summary node: Add an AI node (Basic LLM Chain or AI Agent) after the form and add credentials for a model from OpenAI, Anthropic, or Google. Spell out the output format in your prompt, for example: "Summarize this inquiry in three lines and label urgency as high, medium, or low."
3. Connect notification nodes: Add a Gmail (or Send Email) node and a Slack node to send the summary and contact details. Drag values from the input panel on the left to drop in data from earlier nodes.
4. Test it: Submit a response through the form's Test URL and run the workflow to check the result at each step.
5. Publish it: Save and publish the workflow so the Production URL goes live. From then on, you'll find responses in the Executions tab rather than in the editor.

Menu and node names shift a little between versions. If your screen looks different, search for the same node in the official n8n docs.

How to connect a model and tools to the AI Agent node

Unlike regular nodes that always run in a fixed order, the AI Agent node looks at each request and lets the AI decide which tool to use. You complete it by attaching three things to the slots underneath: a model, memory, and tools.

1. Pick a trigger: To test it conversationally, use the chat trigger (When chat message received). To run it from a form or a schedule, place the AI Agent node after your existing trigger.
2. Connect a Chat Model: Choose one such as OpenAI Chat Model or Anthropic Chat Model and create credentials with your API key. Start with a lighter, cheaper model to confirm the flow, then upgrade.
3. Connect Memory: If the agent needs to remember earlier messages, attach a memory node like Simple Memory. For one-off summaries, you can skip it.
4. Connect Tools: According to the official docs, you must connect at least one tool sub-node to the AI Agent node. Add tools like a sheet lookup, sending email, HTTP requests, or an MCP server, and write a one-sentence description of when each tool should be used so the AI picks the right one.
5. Write instructions: Put the agent's role, rules, and answer format in its system message, then test it with several questions in the chat panel.

Costs, spending limits, and personal data: what to check first

n8n and your AI model bill you separately. n8n charges per execution, while models from OpenAI or Anthropic charge by the amount of text processed (tokens), so set limits on both to avoid surprise bills.

• Set model spending limits: In your AI provider's billing settings, set a monthly cap and usage alerts first. A looping workflow that keeps resending the same request after an error can run up costs fast.
• Watch execution counts: A schedule trigger that fires every few minutes can run hundreds of times a day. Set only the frequency you need and check counts regularly in Executions.
• Minimize personal data: Don't pass names, phone numbers, or ID numbers to the AI unless the summary truly needs them. Pass only the fields the next node requires.
• Keep keys in Credentials: Never paste API keys into prompts or node notes. Store them only in the Credentials menu so they don't travel along when you share or export a workflow.
• Check company policy: Confirm whether customer data may be sent to outside AI services, and consider self-hosting if needed.

Use meeting notes from daglo as the starting point for automation

The best source of automation ideas is the requests that keep coming up in meetings. When you transcribe meetings with daglo, you get a transcript that shows who said what, plus a summary, which makes it easy to spot the tasks people still do by hand every week.

1. Get everyone's consent, record the meeting, and transcribe it in daglo.
2. Ask Board Chat: "List the manual tasks that came up repeatedly in these meetings, with each task's input and output, as a table." Then check the answer against the original transcript right away.
3. Paste the summary into an n8n form or a sheet, and build a workflow that takes it as input to send alerts to the owner or draft a follow-up email.

If you want every meeting summarized as "repeated tasks and automation ideas," save that request in daglo's My Templates. daglo remembers the prompt and automatically summarizes each recording in that format as soon as it ends.

In short, the fastest way to learn n8n is to start with one small automation like form trigger → AI summary → alert, confirm it in Executions, and then expand with the AI Agent node and tools once you're comfortable.

#n8n #HowToUseN8n #WorkflowAutomation #NoCodeAutomation #AIAgents #Zapier #Make #daglo

👉 Start transcribing meetings for free