AI automation for business: 15 practical examples that work
15 practical AI automation examples for business: leads, invoices, reviews and customer service, with the n8n steps, the AI role and where humans decide.
Readn8n automation
An n8n workflow pulls contacts from a Google Sheet, writes each person their own message, sends the email and, for mobile numbers, an SMS too. The sheet shows you straight away what went out and what was delivered.
In short
Automated, AI-personalised email and SMS outreach is an n8n workflow that, on a schedule, takes the next batch of contacts from a Google Sheet and writes a separate message for each one, using their name, context and an offer matched to what the sheet says about them. It then sends the email and, if the number is a mobile, an SMS through a gateway API, and checks whether the text was delivered. It suits businesses that write to customers or leads on a regular basis: follow-ups after meetings, reminders, invitations, offers to their own list. The sheet stays your control panel, showing who got an email, who got a text and what happened to each.
Workflow
A simplified, interactive version of our workflow. Drag the nodes and run it to watch the data flow.
Drag the nodes · click “Run”
Execution log
AI email + SMS: Sheet → AI writes the message → email → number check → SMS via gateway → statuses back in the sheet
Screenshot from the build
A real, unedited screenshot from the n8n editor. The interactive version above is simplified; here you can see every node, branch and error path.

Sound familiar?
One-template bulk email looks like spam, and gets read that way.
The language model writes every message individually from the columns in your sheet: name, industry, last contact, product. An output parser makes sure there is always a subject and a body in a fixed format.
You send texts from a separate dashboard and never know which ones arrived.
The workflow checks whether the number is a mobile, sends the SMS through the gateway API and keeps asking for the status until the message leaves the queue. The outcome (delivered, rejected, failed) lands in the sheet in plain words.
Sending to the whole list at once runs into mailbox limits or a block.
Contacts go out in small batches with pauses between messages. Each row gets a status as soon as it is sent, so after a pause or an error the workflow never writes to the same person twice.
How it works
A Schedule Trigger runs the workflow at set times, for example on weekday mornings. For testing you can start it by hand.
Only rows waiting to be sent are read from Google Sheets, and a Limit node caps how many go in one run. A loop then handles them one by one.
A Basic LLM Chain, using a model via OpenRouter or OpenAI, writes the subject and body from the contact’s data and your guidelines: tone, length, sign-off and what must never be promised.
The email is sent from your own mailbox and the row gets a status and date. A short wait spreads the sending out over time.
A Code node cleans up the number format and an IF node lets only mobiles through. Landlines and invalid numbers are flagged in the sheet.
An HTTP Request sends the text through the gateway API. If the gateway rejects it, the workflow logs the reason and moves on to the next person.
The workflow waits a moment and asks the gateway for the status while the SMS is still queued. It turns the status code into a readable label and saves it next to the contact.
This workflow is built for regular, personal contact with your list, not for a designed newsletter. Instead of a campaign tool you have a sheet: each row is a person and the columns describe what you know about them. AI uses that data to write a message that reads like a normal email from a human, not a template with the first name swapped in.
How the AI writes. The prompt holds your rules: tone, length, sign-off, off-limits topics and anything the message must never promise, such as discounts or dates that aren’t in the sheet. The model only sees the columns it needs for personalisation. A Structured Output Parser enforces the response format, so an empty or cut-off message never reaches the send step.
A human at the start. When we deploy it, the first runs are in test mode: messages go to your own address or wait in the sheet for approval. Once you’re happy with the copy, you flip a status in the sheet and sending to customers begins. You can keep approval switched on for selected segments, such as your biggest accounts.
Statuses, retries and errors. Every row has a separate status for email and SMS. The workflow tells apart a non-mobile number, a gateway rejection, a queued message and a delivered one. Status polling has a retry cap, so the loop never spins forever. A mailbox or API error doesn’t stop the whole batch: the row gets a description of the problem and you get an alert.
Consent and opt-outs. We can set the workflow to write only to rows marked as consented and to skip anyone who has unsubscribed. The legal basis for contacting people is yours to own, but the automation applies the rule every single time, with no quick exceptions.
New to the tool? Read what n8n is and how it works. For more ideas on automating customer communication, see our 15 AI automation examples for business, or get in touch to talk through your setup.
FAQ
Not if you write to people who agreed to hear from you and you have something specific to tell them. The workflow doesn’t buy lists or scrape addresses; it only works with your sheet. Personalisation and a steady pace make the messages look like ordinary correspondence, but consent always comes first.
The workflow uses models through OpenRouter or directly from OpenAI, so you can switch models without a rebuild. For short messages a cheaper model is usually enough. You pay the provider for usage from your own account, and the cost depends on how many messages you send and how long they are.
Any gateway with an API for sending messages and checking their status. The SMS goes out through an HTTP Request node, so switching gateways usually means a new API address, new credentials and a different mapping of status codes.
Yes. The simplest option is for the AI to save the copy in the sheet, with sending starting only once you change the status to “approved”. You can also get a preview on Slack or Telegram with an approve button.
Once you have the contact sheet, a mailbox and an SMS gateway account, setup with testing usually takes 1–2 weeks. Most of that time goes into tuning the prompt on your real examples so the messages sound like you.
Insights

15 practical AI automation examples for business: leads, invoices, reviews and customer service, with the n8n steps, the AI role and where humans decide.
Read
n8n is an AI-ready automation platform you can self-host. Learn how it works, what it costs, what its licence allows and how to build your first workflows.
Read
n8n, Make or Zapier? We compare pricing (checked November 2026), how each meters usage, AI agents and self-hosting to help you pick the right platform.
ReadGet a quote
Describe your process and the tools you use. We reply within one business day with an implementation plan.
Other builds
We use essential cookies and, with your consent, analytics (Google Analytics, Hotjar) and marketing cookies to understand how you use the site and improve it. Details in our privacy policy.