Fundamentals of n8n Workflow Automation
Understand the mental model, watch a complete workflow come together live, and take home a starter library of workflows you can import and run.
Aemal Sayer
CTO & Co-Founder · Avanai
Avanai is an n8n expert partner, delivering hands-on n8n and AI-automation enablement. I help teams go from zero to building real automations - exactly like the AI agent you'll build today.
One AI agent, from chat to answer
First the fundamentals, then I build exactly this live in n8n: an AI agent you chat with that decides when to pull Jira issues, remembers the conversation, and answers in plain language. Here is what it looks like on the n8n canvas.
Follow every click, live, on your own screen.
AvaStage mirrors my screen to your device, live. Rewind any step you miss, and fire your questions in as we go - don't save them for the end.
Point your phone camera here
Triggers: what starts a flow
- One or more triggers, each a possible entry point.
- Manually, on a schedule, a webhook, a form, or a chat message.
- Without a trigger, nothing happens.
Actions: what the flow does
- Any number of steps, one after another.
- Fetch, transform, send, create data.
- Each node does exactly one thing.
Credentials: enter once, reuse everywhere
A credential is where a node stores its connection. You enter the token once, then every node reuses it by name.
Paste the app's token into the credential a single time.
Any node picks the credential from a dropdown.
Stored encrypted, never inside the workflow. Safe to share.
The canvas: where the workflow takes shape
The canvas is your visual workspace. You drop nodes on it, wire them together, and read the whole automation left to right - from the trigger to the result.
One open workspace. Drag nodes on, arrange them, and see the whole flow at a glance.
Each node is one step - a trigger or an action. It does exactly one thing.
Arrows wire nodes in order. Data flows from the trigger down the chain to each next step.
Open a node: input, settings, output
Click any node to open it. It always has the same three parts, left to right - what comes in, what you change, and what goes out.
The data arriving here is the output of the previous node - here, the Hourly Trigger.
Where you configure the node - it enriches or changes the data coming from the input.
The result after your settings run - this is what passes on to the next node.
Those numbers are items flowing between nodes
Look at the labels on the arrows - 1 item, 24 items, 10 items. Each one is how many items pass to the next node, and every node runs once for each item it gets.
The count on each connection is what the next node receives - 1 in, then 24 out of "Get Jira Tickets".
A node runs once for every item it receives. 24 items in means it does its job 24 times.
Nodes reshape it: a fetch returns 24, "Limit" trims to 10, a Loop hands over 1 at a time.
Every run is recorded and replayable
Open the Executions tab to see every time the workflow ran. Each run is a saved snapshot - you can reopen it, see exactly what happened, and reference it by its unique ID.
Manual or scheduled, each execution is logged here with its status, start time and duration.
Every run gets its own ID - open it, share it, or reference it when you're debugging.
It captures the exact data from that moment - inspect what each node received and returned.
Build your first AI Agent
You hand it a goal - it chats with you, decides when to pull Jira issues, and answers in plain language.
- 🧠 Chat model - the reasoning brain (Claude on Bedrock).
- 💬 Simple Memory - remembers the conversation.
- 🔧 Jira tool - fetches issues on demand.
