AI Automation for SMEs: 5 Workflows That Pay Off

AI automation for small businesses in Switzerland: five workflows, what each needs, where a person decides and what data protection law requires.

AI & AutomationTamino HaasPublished: 8 min read
Two people at a wooden table with laptops and notes, planning a workflow.

AI automation pays off for an SME wherever the same work comes up every day: sorting emails, scheduling appointments, logging enquiries, typing up receipts and answering standard questions. This post walks through five of these workflows: what each one needs, where a person has to make the call and what the Swiss Federal Data Protection Act (FADP/DSG) requires.

How widespread is AI in Swiss SMEs?

AI has arrived in Swiss SMEs, but mostly without clear ground rules. According to the AXA SME Labour Market Study 2025, 34% of Swiss SMEs have built AI into their day-to-day work, up from 22% the year before (SECO’s SME portal, in German, 5 November 2025). Businesses use AI most often for translation (52%) and correspondence (47%), and 34% already use it to automate individual work steps (same source).

The gap is in the rules: only just over a third of SMEs have set out which data employees may enter into AI tools, and among businesses with fewer than ten staff the figure is 23% (same source). In the EU, some companies have looked at AI but not adopted it. Their main reasons are a lack of expertise (71%), unclear legal consequences (53%) and data protection concerns (49%), according to the 2025 survey by Eurostat. The five workflows below are small enough to get started with, and each one shows what you need to watch for on data protection.

The five workflows at a glance

Workflow Typical tools Where a person decides Data protection focus
Pre-sorting emails Microsoft 365 or Google Workspace, AI assistant Sending every reply Mailbox contents, storage location
Booking appointments Online calendar, booking tool, chat or phone assistant Special cases, cancellations Contact details, reminders
Enquiries into the CRM Form, CRM, Make, Zapier or n8n Priority, first contact Prospect data
Reading receipts Receipt inbox, text recognition, accounting Every payment Supplier and bank details
Chatbot for common questions Knowledge base, chat widget Advice, complaints Labelling as AI

1. Pre-sort emails and prepare replies

The AI reads incoming emails, sorts them into categories such as enquiry, invoice or support, and drafts a reply. You or your team check the draft and send it. Correspondence is the second most common use of AI in Swiss SMEs, so for many this is the obvious place to start.

What it needs: access to the mailbox, five to eight clear categories with two or three sample replies each, and the rule that nothing leaves the business without a person signing it off. If your mailbox isn’t in the cloud yet, that’s the first step; more on this under IT & Cloud.

Where a person decides: complaints, quotes and anything with legal effect. Here the AI saves you typing, not responsibility.

2. Let appointments book themselves

A booking workflow answers ‘When are you free?’ without the back and forth. Customers pick a free slot online, the calendar is blocked, and the confirmation and reminder go out automatically. AI comes in when requests arrive in free text: a chat or phone assistant understands ‘next Tuesday afternoon’ and suggests suitable free times. The foundation is always a solid booking system, such as WakeSessions, the tool we built for Lohr Wassersport that lets guests reserve their wakeboard sessions online.

What it needs: a well-kept calendar with opening hours and buffer times, a list of services with their duration, and a way to reach a person for special cases, such as a phone number in the confirmation.

Where a person decides: last-minute cancellations, emergencies and anything that doesn’t fit the standard services.

3. Send enquiries straight into the CRM

Every enquiry that comes in through the website, email or chat lands in the CRM automatically, gets tagged with company and topic, and is given a next task, for example ‘Quote by Friday’. That way no enquiry gets lost in the inbox. The AI reads free-text messages, picks out the company, the request and how urgent it is, and suggests the next task. 20% of Swiss SMEs already use AI in customer relationship management (SECO’s SME portal).

What it needs: a CRM, even a simple one, a contact form with clear fields and a connection between the two, for example with Make, Zapier or self-hosted n8n. If you self-host, the data stays on a server of your choice.

Where a person decides: judging which enquiry takes priority, and making the first personal contact.

4. Read receipts and documents

Invoices, delivery notes and forms arrive as PDFs or photos. The AI reads the supplier, date, amount and VAT and passes the data to your accounting. Instead of typing everything up, you only check it.

What it needs: a fixed inbox for receipts, such as a dedicated email address, a mapping of the fields to your accounting software, and a review step: if the AI can’t read a field with confidence, the receipt comes to you for checking.

Where a person decides: every payment. The AI captures the data, you approve it.

5. Answer common questions with a chatbot

A chatbot on your website answers questions about opening hours, prices, processes and services around the clock. It only answers from your own knowledge base and refers people to your team when it’s unsure. Our page on AI chatbots and automation shows how we build these assistants.

What it needs: a well-kept knowledge base with common questions, prices and terms, a visible notice that an AI is answering, and a handover to a person by form, email or phone.

Where a person decides: advice, complaints and anything personal.

What does data protection law require for AI?

The FADP applies directly to AI as soon as personal data is involved. The Federal Data Protection and Information Commissioner, FDPIC (EDÖB), states that the technology-neutral law in force since 1 September 2023 applies directly to AI-supported data processing (FDPIC, in German, updated 8 May 2025). For day-to-day SME work, that means four things:

  1. Transparency: anyone talking to a chatbot has the right to know that they are communicating with a machine and whether their input is used to improve the system (FDPIC, same source). Put this in your privacy policy as well.
  2. Storage location and contract: find out where a tool processes the data, and sign a data processing agreement with the provider. If data is disclosed abroad, for example to a provider in the US, additional rules apply.
  3. Internal rules: set out in writing which data may go into the tool. Health data, personnel files and passwords don’t belong in free chat tools.
  4. Human review: if a system makes an automated decision that has legal consequences for a person or significantly affects them, that person can ask for a human to review it (FDPIC, same source).

If you serve customers in the EU, you should also know the EU AI Act. Its Article 50 has applied since 2 August 2026 and requires providers of AI systems that interact with people to make that clear, at the latest at the first interaction (Regulation (EU) 2024/1689, Article 50). A clear notice in the chat window saying that an AI is answering, and whether input is reused, meets the core of the transparency obligations under both sets of rules.

Where you’re better off not automating

Not every workflow is a good fit. Leave automation out when:

  • a mistake would be costly or sensitive and nobody checks, for example with payment reminders or notices of termination
  • the workflow rarely comes up; then it isn’t worth setting up
  • the data is incomplete or scattered; then tidy it up first
  • the personal relationship is the product, as with consulting

How should you get started?

Start with the workflow that eats up the most time today and follows clear rules. For most SMEs, that’s pre-sorting emails or booking appointments.

  1. Pick a workflow: for one week, note down which tasks repeat and how long they take.
  2. Set the rules: what is the AI allowed to do, what isn’t it, and who checks?
  3. Test small: two to four weeks with real cases, but with a person signing off.
  4. Measure: compare time saved and error rate against your list from step 1.
  5. Expand: only move on to the next workflow once the first one runs reliably.

We’re happy to support you with the rollout, from choosing the right workflow to running it in line with data protection law. For a no-obligation first conversation, get in touch via our contact page.

Frequently asked questions

Is AI automation worth it for very small businesses?

Yes, especially there. In a business of three to ten people, routine work falls on just a few shoulders, so every hour you automate makes a noticeable difference. What matters is a clearly defined workflow rather than a big project.

Can I enter customer data into ChatGPT and similar tools?

Only if you know where the data is processed, a data processing agreement is in place and your customers have been informed. For free versions without such commitments, the rule is simple: don’t enter any personal data.

Do I have to label my chatbot as AI?

Yes. In the FDPIC’s view, users have the right to know that they are talking to a machine. For services aimed at people in the EU, Article 50 of the EU AI Act also requires it.

Which workflow should I start with?

The most frequent one with the clearest rules. For most SMEs, that’s pre-sorting emails or booking appointments, because both save time quickly and a person can easily double-check the results.

Topics

  • #AI
  • #Automation
  • #SMEs
  • #Data protection

Ready for a new look?

Tell us in a few sentences what you have in mind. You get a reply within 24 hours and a date for the first call.

We only use your details to answer your request. Privacy · Tell us about your project

Cookie settings

Decide which services we may load. You can change your choice at any time via “Cookie settings” in the footer.

Cookies · Privacy