What Can You Automate? 25 Examples From Small and Mid-Sized Businesses

You can automate any process that has a clear trigger, follows fixed rules and works with data that already sits in a system. That applies to more work than most managing directors expect, and to less than the advertising promises.

"What could we actually automate?" is a question we hear again and again in first conversations. The most honest answer is a list. Not because you need all 25 examples, but because while reading you will stop at three or four and think: "We do exactly that, by hand." Those spots are where you start. Once the first numbers for them are on the table in that first conversation, the reaction we hear most often is: "Wow, okay, we're losing that much time right now? I'd never have guessed that." Before the list comes a short check with three questions you can run on any process of your own, and after the list the question of what to begin with.

How do you know whether a process can be automated?

Three questions are enough for a first assessment. If you answer all three with yes, the process can very likely be automated:

Question	Yes means	No means
1. Is there a clear trigger?	An email arrives, a form is submitted, a date is reached, a status changes	The process starts "when someone remembers"
2. Does it follow rules you can write down?	If A then B; approval above amount X; special price for customer Y	Every case is decided fresh, without a pattern
3. Does the data sit in a system?	CRM, ERP, accounting, calendar, mailbox, spreadsheet	The information only exists in one person's head

Question 2 has changed in recent years. A few years ago every rule had to be spelled out exactly. Now AI agents take over the part that has to read, classify or summarise text, such as "Is this email an invoice, an order or a complaint?". That moves many processes that used to fail question 2 into the "yes" column. How to find the processes that cost you the most time is covered in Find Your Time Sinks.

25 examples: what can be automated in five areas

Each example names the trigger, what happens automatically afterwards and which type of tool typically handles it: an automation platform such as Make.com or n8n, an AI agent, or a small piece of custom software, the three building blocks we describe on our services page. All 25 are patterns as they occur in companies with 10 to 100 people. For one area only, we quote a figure from a real client project below.

Back office and bookkeeping

#	Trigger	What happens automatically	Tool
1	An invoice arrives by email	Attachment recognised, amount, date and sender extracted, invoice recorded in the accounting system	AI agent + platform
2	A payment arrives in the bank account	Payment matched to the right invoice, open items updated, exceptions queued for review	Platform
3	Month end	Open invoices collected, payment reminders drafted, list sent to bookkeeping	Platform
4	A new supplier is created	VAT ID checked, master data created in ERP and accounting, filing folder set up	Platform
5	An employee submits expenses	Receipt read, category suggested, approval routed to the manager, booking prepared	AI agent + platform

The core of examples 1 and 2 runs in our own business: receipts are recognised in the mailbox, created in the accounting system and matched to the right bank transaction. That is how our own bookkeeping is around 90 percent automated, while our obligations under GoBD (the German rules for proper digital bookkeeping) and our tax adviser's review stay exactly where they were. What that means in detail and where the limits are is in Automate Your Bookkeeping, and how the inbox gets sorted before an invoice is even recognised is in Automate Your Inbox.

Sales and quotes

#	Trigger	What happens automatically	Tool
6	An enquiry comes in through the contact form	Enquiry classified, created in the CRM, assigned to the right salesperson, acknowledgement sent	AI agent + platform
7	A salesperson selects product, quantity and customer	Prices, discount tiers and surcharges calculated, quote generated as a PDF	Custom software
8	A quote has been unanswered for 7 days	Follow-up email drafted, task created in the CRM, salesperson notified	Platform
9	A salesperson records a voice note after a customer meeting	Note transcribed, summarised, entered in the CRM, next steps created as tasks	AI agent
10	An order is confirmed	Order created in the ERP, invoicing data carried over, project folder created, team informed	Platform

Example 7 is the classic that grows out of an overloaded spreadsheet with dozens of tabs. How such a workflow is built is shown in Automated Quote Generation.

Customer service and order handling

#	Trigger	What happens automatically	Tool
11	A customer emails support	Request classified (question, complaint, order), ticket created, standard questions answered as a draft	AI agent
12	A customer asks "Where is my order?"	Status read from the inventory system and prepared as a reply, without anyone having to search the system	AI agent
13	An order comes in through the shop	Data checked, handed to inventory and shipping, exceptions queued for approval	Platform or custom software
14	A complaint is logged	Return label created, replacement or credit note prepared, customer informed, case queued for approval	Platform
15	A delivery is completed	Customer receives confirmation, a review request after seven days, a ticket if there is a problem	Platform

Examples 11 and 12 are often worth a look first, because the same questions get answered anew every day. What an AI agent can really take on in support and what stays with people is in Automate Customer Support.

Marketing and partners

#	Trigger	What happens automatically	Tool
16	A new blog article is published	Summary, social media drafts and newsletter paragraph created and queued for approval	AI agent
17	A new newsletter subscriber	Welcome series starts, contact created in the CRM, source recorded	Platform
18	Start of the month	Website, ad and newsletter figures collected, compared and summarised as a plain-language report	AI agent + platform
19	A partner signs up for an affiliate programme	Check, activation, login details, onboarding email, entry in the partner list	Platform
20	A customer leaves a review	Review detected, reply drafted, critical ones flagged to the person responsible	AI agent

From a client project in affiliate operations: managing more than 800 partnerships went from around 30 hours a week to about 15 minutes. How AI agents take over recurring reports like example 18 is covered in AI Agents for SMEs.

HR and internal processes

#	Trigger	What happens automatically	Tool
21	An application arrives	Documents summarised, acknowledgement sent, interview slot proposed to the department	AI agent + platform
22	A new employee signs	Accounts created, equipment ordered, onboarding plan with dates set up	Platform
23	A holiday request is submitted	Cover checked, approval routed to the manager, calendar and absence list updated	Platform
24	An employee asks about an internal rule	Answer from the handbook stored as the knowledge base, with a reference to the source	AI agent
25	Weekly meeting	Open tasks gathered from all tools and sent to the team as a weekly overview	Platform

These five are often underrated because they do not bring in revenue directly. But they cost the same hours every week, and several of them meet all three criteria from the check above.

Where should you start?

Not with the biggest example, but with the one that has three properties: it happens often, it follows clear rules, and a mistake is not expensive. Often that is the inbox or receipt capture. Start there and you have a result the team can see within a few weeks, and then the second example comes easier than the first. How such a start runs as a sprint is on our services page.

Two warnings from our own practice. First, do not automate a process you have not tidied up beforehand. A chaotic process only becomes faster chaos. Second, for anything that leaves the building or moves money, have a person approve it in the first weeks. The approval costs seconds, a wrongly sent quote costs a customer.

Frequently Asked Questions

What can a small business automate?

The same processes as a large one, just with smaller tools. Often an automation platform, with an AI agent where needed, that connects mailbox, accounting and CRM is enough. Examples 1, 3, 6, 8 and 23 are usually the quickest to set up in small businesses, because they connect only a few programs. What matters is that the programs involved can be connected to other software, and most common tools offer that today.

What cannot be automated?

Anything that needs a judgement that does not fit into rules: negotiations, staffing decisions, creative work, dealing with an angry customer. Processes whose data only lives in one person's head cannot be automated either, as long as that stays the case. Automation takes over preparing, sorting and forwarding. The decision stays with the person.

Do I need custom software for automation?

Usually not. Most of the 25 examples run on an automation platform such as Make.com or n8n, which connects existing programs through their APIs, or on an AI agent that looks things up in your systems. It only becomes custom software when a process needs its own screen or a system cannot be connected to other software, as with the quote calculator in example 7. Which of the three fits when is on our services page.

What is the difference between automation and an AI agent?

An automation follows fixed rules: when an order comes in, create it in the ERP. An AI agent can additionally read, classify and write: it recognises that an email is a complaint and drafts a reply. In practice the two work together. The agent understands the case, the automation handles the steps that follow. More on that in AI Agents for SMEs.

How much time does automation really save?

That depends on how often the process occurs and how many steps it has. A process that costs three minutes ten times a day gives you half an hour back daily. A monthly report that takes an afternoon gives you an afternoon a month. That is why it pays to measure your own time sinks before the first project instead of guessing.

The next step

If you stopped at three or four examples while reading and thought "we do exactly that, by hand", this is the moment to get your own numbers. That is what we do in the free Time Potential Analysis: 45 minutes, no sales pitch, and you leave with your three biggest time sinks, the savings potential in hours and a recommendation on where to start. Even if you never work with us afterwards.