Most SMEs are sitting on a significant productivity opportunity they have not yet touched. Their teams spend hours every week on work that follows predictable rules, routing emails, extracting data from documents, chasing approvals, updating records across multiple systems, generating reports from the same sources in the same format. None of this requires human judgement. All of it can be automated.
n8n has emerged as one of the most powerful tools available for exactly this kind of automation, and unlike most enterprise automation platforms, it is genuinely accessible to small and medium businesses, particularly when self-hosted. What makes n8n stand apart in 2026 is not just its workflow automation capabilities, but its native support for AI agents: autonomous workflows that can reason, make decisions, call external tools, and loop until a task is complete, without a human in the loop.
This article covers what n8n AI agents are and how they work, six practical use cases for SMEs with realistic estimates of time saved, a straight comparison of n8n Cloud versus self-hosting, and a clear step-by-step guide to setting up your own self-hosted n8n instance.
What Is n8n, and What Makes It Different?
n8n is an open-source workflow automation platform. Like Zapier or Make.com, it lets you connect applications and automate processes through a visual drag-and-drop editor, no application code required. But n8n goes further in several important ways:
- It is open source and self-hostable, you can run it on your own server, giving you full data control and unlimited executions at the cost of a VPS (typically €10–20/month)
- It supports native AI agents, not just "call OpenAI and do something with the result", but true autonomous agent loops where n8n gives an AI model access to tools and lets it decide which ones to use and in what order
- It has deep data transformation capabilities, built-in JavaScript/Python nodes, JSONata expressions, and structured data manipulation that most no-code tools cannot match
- It connects to over 400 integrations natively, plus any API via HTTP nodes
- It runs complex multi-branch, looping, and parallel workflows that simpler tools struggle to model
The result is a platform that sits between a traditional no-code tool and a custom-built automation solution, powerful enough for genuinely complex business logic, but accessible enough that a technically minded business owner or operations manager can build and maintain workflows without a development background.
What Is an n8n AI Agent?
An n8n AI agent is a workflow that gives an AI model, typically GPT-4o, Claude 3.5, or a local model like Llama, access to a set of tools, and lets the model decide autonomously how to use those tools to complete a task.
This is different from a simple "send data to AI and get text back" automation. In a traditional AI node, you pass in some text, the model responds, and the workflow continues. In an agent loop, the model can:
- Read from a database or spreadsheet to gather context
- Search the web or an internal knowledge base for relevant information
- Send an email, create a calendar event, or update a CRM record
- Check the result of its own action and decide whether to continue or stop
- Escalate to a human if it encounters a situation outside its confidence threshold
The model runs in a loop, choosing tools, executing actions, observing outcomes, until it determines the task is done or needs human input. This is what makes AI agents fundamentally different from simple automation: they handle situations with variability that a rigid if-then workflow cannot anticipate.
n8n was one of the first automation platforms to implement the ReAct agent loop natively, the architecture that allows a language model to reason, act, and observe in cycles. Combined with self-hosting and unlimited executions, this makes n8n the most cost-effective platform for running AI agent workflows at any volume. Enterprise-grade AI automation for the cost of a monthly VPS.
Every n8n plan, including the free Community self-hosted version, includes access to AI agent nodes, memory nodes, and tool integrations. There are no AI feature paywalls.
6 Practical n8n AI Agent Use Cases for SMEs
The following use cases are drawn from patterns we implement with Irish and international SMEs. For each, we include a realistic estimate of time saved per week based on typical team sizes and process volumes, as well as the complexity of the n8n workflow involved.
Email Triage and Intelligent Routing
A shared inbox, whether a sales@, info@, or support@ address, generates a constant stream of incoming emails that require someone to read, categorise, and route to the right person or system. Most businesses handle this manually, often with a dedicated person spending 1–2 hours every morning working through overnight messages.
An n8n AI agent connected to Gmail or Outlook reads each incoming email, classifies it (sales enquiry, support request, invoice, complaint, supplier communication, spam), extracts key data (sender, urgency, product mentioned, account reference), and takes the appropriate action: creating a CRM lead, opening a support ticket, forwarding to the relevant team member with a summary, or flagging for urgent review.
The agent can also draft a personalised acknowledgement reply, not a generic auto-reply, but a contextual response that references what the sender asked about, and send it immediately, often before a human would even have read the email.
per team member
Invoice Processing and Accounts Payable
For most SMEs, invoice processing is a deceptively expensive process. Someone opens the email, downloads the PDF, reads it, manually types the supplier name, invoice number, line items, and amounts into an accounting system (Xero, QuickBooks, or Sage), and files the document. Multiply this by 30–100 invoices per month and you are looking at hours of work that adds zero business value.
An n8n workflow monitors an inbox for incoming invoices, extracts the PDF attachment, passes it through a document AI node (using OpenAI's vision API or a dedicated tool like Google Document AI), extracts all structured data, supplier, date, line items, VAT, total, payment terms, validates the result against expected suppliers, and creates a draft bill directly in your accounting software via API. The original PDF is filed automatically.
Exception handling is built in: if the AI confidence is below a threshold, or the supplier is not recognised, the workflow pauses and notifies an accountant to review. Everything else is fully hands-free.
in finance admin
Sales Lead Research and Personalised Outreach
When a new lead enters your CRM, from a web form, LinkedIn, a trade show list, or a referral, someone typically spends 15–30 minutes researching the company, understanding what they do, finding the right contact, and drafting a personalised first email. For a team generating 20–40 new leads per week, this is a significant time drain that also suffers from inconsistency.
An n8n AI agent triggered by a new CRM contact automatically searches the company website and LinkedIn, extracts key business information (industry, size, recent news, technology stack, likely pain points), cross-references this against your service offering, and drafts a personalised outreach email that references something specific about the prospect's situation. The draft is placed in a human review queue, the sales rep reads, approves with one click, and sends. Research time drops from 25 minutes to under 2 minutes per lead.
The same agent can enrich the CRM record automatically, adding company size, industry, website, LinkedIn URL, so your pipeline data is always complete without any manual data entry.
per sales person
Internal Knowledge Base and Employee Q&A Agent
Every growing SME accumulates internal knowledge that lives in scattered places, onboarding documents in Google Drive, HR policies in a shared folder, process guides in Notion, pricing details in a spreadsheet. When an employee needs an answer, they either ask a colleague (interrupting someone else's work) or spend time searching through folders hoping to find the right document.
An n8n AI agent can be connected to a Slack or Teams channel and given access to your internal document store (Google Drive, Notion, SharePoint) via vector embedding. When a team member asks a question in Slack, "What is our policy on taking time off in lieu?" or "What is the onboarding checklist for a new supplier?", the agent searches the knowledge base, finds the relevant document sections, synthesises a clear answer, and responds directly in the thread with a link to the source document.
The agent can also be trained to know when it does not have enough information and escalate to a manager rather than fabricating an answer, a crucial feature for anything compliance or HR related.
across the team
Weekly Reporting and KPI Summaries
Most SMEs produce weekly or monthly management reports by hand: someone opens multiple systems, exports data into spreadsheets, reformats tables, writes commentary, and emails a PDF to the management team. This process takes 3–5 hours each time and often produces reports that are already slightly out of date by the time they are read.
An n8n workflow triggered on a schedule pulls data directly from your operational systems, your CRM for sales pipeline and win rates, your accounting software for revenue, outstanding invoices and cash flow, your operations platform for job completion rates or delivery performance, transforms it into clean structured data, runs it through an AI node to generate written commentary highlighting key trends and anomalies, and assembles the full report as a formatted email or PDF. The report lands in inboxes automatically every Monday morning at 8am, with no human involvement.
The AI commentary node is particularly powerful here: rather than just presenting numbers, it can flag that "revenue is up 12% month-on-month but the number of outstanding invoices has increased significantly, this may indicate a collections issue worth reviewing."
in reporting
New Employee Onboarding Automation
Onboarding a new team member involves a long sequence of administrative steps that often falls on an office manager or HR person: creating accounts, assigning licences, sending welcome emails, scheduling introductory meetings, collecting documents, setting up payroll, and ensuring the new hire has everything they need on day one. Across a typical SME, this process can consume 5–8 hours of administrative time per new hire, much of it copy-pasting information between systems.
An n8n workflow triggered by a new employee record in your HR system (BambooHR, Personio, or a simple Airtable database) orchestrates the entire sequence automatically: provisioning a Google Workspace account, sending IT setup instructions, scheduling a day-one welcome call in Google Calendar, adding the new hire to relevant Slack channels, creating a personalised onboarding checklist in Notion, and notifying each department lead that a new colleague is joining their area. A follow-up workflow checks in at day 3, day 7, and day 30 to ensure the checklist items have been completed.
new hire
Time Saved: Summary
| Use Case | Hours Saved / Week | Est. Annual Value* | Complexity |
|---|---|---|---|
| Email Triage & Routing | 4–7 hrs | €6,000–€10,500 | Medium |
| Invoice Processing (AP) | 3–6 hrs | €4,500–€9,000 | Medium–High |
| Sales Lead Research & Outreach | 5–10 hrs | €7,500–€15,000 | Medium |
| Internal Knowledge Q&A Agent | 2–4 hrs | €3,000–€6,000 | Medium–High |
| Weekly KPI Reporting | 3–5 hrs | €4,500–€7,500 | Medium |
| Employee Onboarding | 5–8 hrs/hire | €1,500–€2,400/hire | Low–Medium |
| Combined (first 5 use cases) | 17–32 hrs/week | €25,500–€48,000/yr |
* Annual value estimated at €30/hour (blended staff cost for typical SME office or operations role). Figures are indicative, actual savings depend on team size, process volume, and current automation maturity.
n8n Cloud vs Self-Hosted: Which Should You Choose?
This is the most important decision when adopting n8n, and it comes down to three factors: your technical capacity, your data privacy requirements, and the volume of workflows you intend to run.
| Factor | n8n Cloud | Self-Hosted |
|---|---|---|
| Monthly Cost | €20–€50/month (Starter/Pro) | €10–€20/month (VPS only) |
| Execution Limits | 2,500/month (Starter) · 10,000/month (Pro) | Unlimited |
| Setup Required | None, sign up and start | ~30–60 mins initial setup |
| Maintenance | Fully managed, zero ops burden | Occasional updates, backup management |
| Data Privacy | Data processed on n8n servers | All data stays on your own server |
| GDPR / Compliance | n8n is EU-based · GDPR compliant | Full control, no third-party data processing |
| Uptime / Reliability | SLA-backed, managed infrastructure | Depends on VPS provider (typically 99.9%) |
| Best For | Non-technical teams · Low–medium workflow volume · Fast start | High-volume workflows · Sensitive data · Cost-conscious teams |
| AI Agent Features | Included on all plans | Included, Community edition is fully featured |
n8n Cloud Pricing (2026)
| Plan | Price (annual) | Executions / Month | Best For |
|---|---|---|---|
| Starter | €20 / month | 2,500 | First workflows · Evaluation |
| Pro | €50 / month | 10,000 | Growing SMEs · Multiple workflows |
| Business | €667 / month | 40,000 | Enterprise teams · High volume |
| Self-Hosted | €10–20 / month | Unlimited | Any SME with basic technical capacity |
The Self-Hosting Cost Argument
If you run even 5,000 workflow executions per month, a modest volume for a busy SME with 4–5 automated processes, you are already beyond the n8n Cloud Starter plan. The Pro plan at €50/month handles 10,000 executions. A VPS from Hetzner or DigitalOcean capable of running n8n comfortably costs €10–15/month and has no execution limit whatsoever.
The break-even point is approximately 20,000 executions per month. Below that, the Cloud Pro plan may be comparable in cost. Above it, self-hosting becomes dramatically cheaper. For a business running high-frequency automations, processing emails, invoices, or real-time data feeds, self-hosting pays for itself within the first month.
How to Self-Host n8n: Step-by-Step Setup Guide
Setting up a self-hosted n8n instance on a VPS takes approximately 30–60 minutes if you follow these steps. You do not need to be a developer, you need to be comfortable running commands in a terminal window. If you have ever used the command line on a Mac or connected to a server via SSH, this is well within reach.
The recommended setup uses Docker Compose on a Linux VPS (Ubuntu 24.04 LTS), with Nginx as a reverse proxy and a free Let's Encrypt SSL certificate. This is the production-grade setup used by the vast majority of n8n self-hosters.
Choose and Provision a VPS
You need a Linux VPS with at least 1 vCPU and 2 GB RAM, though 2 vCPU / 4 GB RAM is recommended for a comfortable production setup. Recommended providers for Irish and European businesses:
- Hetzner Cloud (Germany), CX22 server at approximately €4.50/month is sufficient for most SME workloads. Excellent value and EU-based.
- DigitalOcean, Basic Droplet at $6/month. Well-documented and widely used for n8n deployments.
- Contabo, Very cost-effective for higher RAM requirements.
- Hostinger VPS, Offers a one-click n8n installer for teams who prefer a GUI approach.
Select Ubuntu 24.04 LTS as your operating system. You will also need a domain name or subdomain pointing to your server's IP address (e.g., n8n.yourbusiness.com), you can set this up in your domain DNS settings with an A record.
Connect to Your Server and Update It
Once your VPS is provisioned, connect via SSH from your terminal (Mac/Linux) or PuTTY (Windows). Replace YOUR_SERVER_IP with the IP address shown in your VPS dashboard:
# Update system packages
sudo apt update && sudo apt upgrade -y
Install Docker and Docker Compose
n8n runs inside a Docker container, this keeps it isolated, makes updates simple, and means you do not need to install Node.js or any dependencies directly on your server:
curl -fsSL https://get.docker.com -o get-docker.sh
sudo sh get-docker.sh
# Install Docker Compose
sudo apt install docker-compose-v2 -y
# Add your user to the docker group (so you don't need sudo each time)
sudo usermod -aG docker ${USER}
newgrp docker
Create the n8n Configuration Files
Create a dedicated folder for your n8n setup and add the two required configuration files:
Create a file named .env with your database credentials and domain. Replace the placeholder values:
POSTGRES_DB=n8n
POSTGRES_USER=n8nuser
POSTGRES_PASSWORD=your_strong_password_here
N8N_HOST=n8n.yourbusiness.com
N8N_PROTOCOL=https
WEBHOOK_URL=https://n8n.yourbusiness.com/
Create a file named docker-compose.yml with the following content:
version: '3.8'
services:
postgres:
image: postgres:16
restart: always
env_file: .env
volumes:
- postgres_data:/var/lib/postgresql/data
n8n:
image: n8nio/n8n
restart: always
ports:
- "5678:5678"
env_file: .env
environment:
- DB_TYPE=postgresdb
- DB_POSTGRESDB_HOST=postgres
- DB_POSTGRESDB_DATABASE=${POSTGRES_DB}
- DB_POSTGRESDB_USER=${POSTGRES_USER}
- DB_POSTGRESDB_PASSWORD=${POSTGRES_PASSWORD}
volumes:
- n8n_data:/home/node/.n8n
depends_on:
- postgres
volumes:
postgres_data:
n8n_data:
Start n8n
Launch the containers with a single command:
# Verify both containers are running
docker compose ps
n8n is now running on port 5678 of your server. You can access it at http://YOUR_SERVER_IP:5678 to verify it is working before setting up the domain and SSL.
Set Up Nginx Reverse Proxy and HTTPS
To access n8n securely via your domain with HTTPS, install Nginx and Certbot (for free SSL from Let's Encrypt):
sudo apt install nginx certbot python3-certbot-nginx -y
# Create an Nginx config for your n8n domain
sudo nano /etc/nginx/sites-available/n8n
Paste the following into the Nginx config file (replace n8n.yourbusiness.com with your actual subdomain):
listen 80;
server_name n8n.yourbusiness.com;
location / {
proxy_pass http://localhost:5678;
proxy_set_header Host $host;
proxy_set_header X-Real-IP $remote_addr;
proxy_http_version 1.1;
proxy_set_header Upgrade $http_upgrade;
proxy_set_header Connection "upgrade";
}
}
sudo ln -s /etc/nginx/sites-available/n8n /etc/nginx/sites-enabled/
sudo nginx -t
sudo systemctl reload nginx
sudo certbot --nginx -d n8n.yourbusiness.com
Certbot will automatically configure HTTPS and set up auto-renewal. Your n8n instance is now accessible at https://n8n.yourbusiness.com with a valid SSL certificate.
First Login and Initial Configuration
Visit your n8n URL in a browser. You will be prompted to create an owner account (email and password). Once logged in:
- Go to Settings → Credentials and add the API keys for the apps you want to connect (OpenAI, Gmail, Slack, etc.)
- Create your first workflow using the + New Workflow button
- Explore the built-in AI Agent node, this is where you configure your AI model, define available tools, and set memory behaviour
- Enable workflow activation to make a workflow run automatically on a schedule or when triggered by an external event
Set Up Automated Backups
This step is essential and often skipped. Your workflows and credentials live in the PostgreSQL database, if your server fails without a backup, you lose everything. Set up an automated daily backup:
nano ~/n8n/backup.sh
# Add this content to the file:
#!/bin/bash
docker compose -f ~/n8n/docker-compose.yml exec -T postgres \
pg_dump -U n8nuser n8n > ~/n8n/backups/n8n_$(date +%Y%m%d).sql
# Make the script executable and schedule it with cron
chmod +x ~/n8n/backup.sh
mkdir ~/n8n/backups
crontab -e
# Add this line to run at 2am daily:
0 2 * * * ~/n8n/backup.sh
For production use, also copy your backups to an off-site location, an S3 bucket, Backblaze B2, or even Google Drive, using an additional step in the backup script.
Updating n8n
Keeping n8n up to date is simple with Docker. Run these commands whenever a new version is released (n8n publishes release notes on their blog):
cd ~/n8n
docker compose pull n8n
docker compose up -d
This process takes under a minute and causes only a brief interruption. Your workflows, credentials, and data are stored in the persistent PostgreSQL volume, nothing is lost during an update.
Is Self-Hosting Right for Your Business?
Self-hosting n8n is the right choice if you meet at least two of the following:
- You have someone in the team (or a consultant) who is comfortable working in a terminal and following technical instructions
- Your workflows will run at high volume, processing hundreds or thousands of records per day
- You handle sensitive data (personal data, financial records, health information) and want to keep it entirely within your own infrastructure
- You are cost-conscious and want to avoid execution limits and scaling costs
- You want to run local AI models (Ollama, LM Studio) rather than sending data to OpenAI, only possible with self-hosted n8n
If your team is fully non-technical and speed of setup is the priority, start with n8n Cloud Starter at €20/month. You can always migrate to self-hosted later, n8n supports exporting all your workflows as JSON files, making migration straightforward.
Getting Started with n8n at Data Sensum
At Data Sensum, we design, build, and deploy n8n automation systems for Irish SMEs, from standalone workflows that solve a single bottleneck to full automation stacks that cover sales, operations, and reporting in a single integrated system. We also set up and maintain self-hosted n8n instances on behalf of clients, handling server configuration, security, backups, and updates so the internal team does not need to.
If you would like to explore what n8n could automate in your business, our free productivity audit is a 45-minute conversation where we map your current manual processes, identify the highest-value automation opportunities, and give you a realistic picture of what is achievable, with no obligation to proceed.
The businesses that move fastest on this are not the ones with the biggest IT budgets. They are the ones who identify one or two high-value processes to automate, prove the value quickly, and then expand. n8n makes that first step very accessible, and the compounding effect on team productivity is significant.