Most Swiss self-employed people don't get stuck with AI because the tools are hard. They get stuck because they can't decide where to begin. There are dozens of small admin tasks, each feels equally urgent, and the quiet worry of picking the wrong one means nothing changes at all.

There is a calm way through this. Before you automate a single thing, spend one ordinary week noticing where your hours actually go. The result is a short, honest list — and from that list, one task stands out clearly as the place to start.

Why the first choice matters more than the tool

It is tempting to begin with the tool. Which model, which subscription, local or cloud? Those questions feel productive, but they are almost never the reason an AI project stalls. The real question is much simpler: which task is worth automating first?

A task is a good first candidate when it passes three tests:

  • It repeats. It comes back every day, week or month, so a routine you set up once keeps paying off.
  • It is low-risk. A draft that still needs your review is fine; a decision that must be right the first time is not a first task.
  • It drains you. It is the task you postpone, the one you quietly resent doing by hand.

The one-week audit, in four quiet steps

You don't need a complicated time-tracking system. A small notebook, a notes app or a simple spreadsheet is enough.

Step 1 — Note the tasks, not the minutes

For five working days, write down every administrative task as it lands: answering routine emails, writing offers, chasing invoices, updating client files, preparing meeting notes. Don't measure minutes — just list what you did and whether you would rather hand it to someone else.

Step 2 — Mark the repeaters

At the end of the week, mark every task that came up more than once. These are your repeaters. A task that appeared five times in one week is worth far more than a rare one-off, even if the one-off took longer.

Step 3 — Score each repeater

Give each repeating task a quick score from one to three for how much you dislike it, how often it returns, and how safe it would be to delegate a draft. Add the three numbers together. The highest total is your shortlist.

Step 4 — Pick one, and only one

Choose the single highest-scoring task and commit to automating just that one for the next month. Not three, not five — one. Starting small is how a new routine actually sticks.

A prompt to find the right one for you

If the list still feels unclear, hand the week's notes to the AI and let it help you rank them:

I am a [notary / fiduciary / doctor / small office] in Switzerland. Here is the list of administrative tasks I did last week and how often each came up: [list]. Help me rank them by (1) how repetitive they are, (2) how low-risk it is to produce a first draft, and (3) how much time they consume. Recommend the single best task to automate first, and explain your reasoning in two sentences.

What a good first task looks like in practice

In a notary's office it is often the intake summary or the first draft of a standard letter. In a fiduciary's practice it is the reminder email or the booking suggestion. For a doctor it is the dictation and the referral letter. In every case the pattern is the same: something that repeats, something a careful human always reviews, and something you will be glad never to type from scratch again.

One honest caveat

This audit is not a productivity contest. You are not measuring yourself against a standard, and the goal is not to squeeze more hours out of every day. You are simply noticing, calmly and honestly, where the friction sits — so that the first step you take is the right one.

The outcome

After one quiet week you have a short list of your real repeaters, a score that points to the right one, and a single, clearly chosen first AI task. You can start with confidence instead of the vague feeling that you should finally do something with AI — and the rest can wait, calmly, until you are ready.