Assessing your AI maturity: what your last ten tasks reveal

Open your calendar and your sent items and go back one week. Pick out ten tasks you actually completed – the most recent ones, not the typical ones. Those ten lines often say more about where you stand with AI than a self-assessment, and they take around twenty minutes.

Anyone setting out to assess their AI maturity – that is, how confidently their work with AI already runs – usually starts with what works every day: the drafts that come together faster, the summaries that land. The calendar shows something else – and that something else is the more interesting half.

Your skill grows where you use it daily

Working with AI regularly produces regular wins. A draft is done in minutes, an outline falls into place, the wording finally works.

Those wins are real. They simply all come from the same slice: a handful of recurring tasks in which you get better week by week. The feeling of "I can do this" is built on repetition – a reliable proof of your ability exactly where you practise.

So this article asks: where has AI not come up for you at all yet?

The third pile

Sort your tasks into three piles for a moment.

The first holds the tasks where you used AI. You know this pile well – it carries your self-image.

The second holds the ones where you thought about it briefly and then decided against it. You remember those too: the decision cost you a moment, and it had a reason.

The third pile is the interesting one. It holds the tasks where the question never came up. You remember this pile least well, and there is a simple reason for that: memory grows out of deliberation, and where none took place, none remains. That is why it hardly comes to mind by itself. Writing it down still works.

And it is usually the largest.

Among those five sat one that returns every month: the commentary accompanying the figures. The numbers themselves came out of the system; the writing around them cost three quarters of an hour each time. It had never occurred to anyone that something might be gained here – because the task had simply been done, for years.

Ten tasks, three marks

The exercise is quick to explain and doable today.

First: write down the last ten tasks. Work backwards through your calendar and sent items and note what you actually completed. Take them as they came, not the ones that seem important – the moment you start considering which tasks are typical, you are back with your self-image.

Second: one of three marks per task. Used · considered · never came up. One mark per line, and no maybes here.

Third: look only at the third group. For each of those tasks, write down the intermediate steps – what it actually consists of. Phrasing, ordering, summarising, drafting: steps like these are often good candidates. Whether they genuinely fit depends further on the content, on data protection, and on how much checking the result needs.

Fourth: take one of them. The one that recurs most often. Next time round, start there.

It works the same with several people. Two or three people each do the exercise on their own, twenty minutes. Put the third piles side by side: whatever shows up on more than one list is the shared first task.

The run takes twenty minutes and delivers something no self-assessment gives you: a list of concrete places instead of a feeling.

The second mark deserves a footnote. Deciding deliberately against AI – because a phone call is faster, because judgement is involved, because the matter is confidential – is a good decision. That pile is a sign of maturity rather than a shortfall.

You can use an AI as a partner for the run-through. It sorts along with you and keeps the record; it is not there to judge.

Your biggest gains often sit in the tasks where AI has simply never crossed your mind.

What this list shows – and where it ends

You now have a list drawn from your own week. It is concrete, it belongs to you, and you can start on it tomorrow.

The list only reaches as far as your week, though: it covers exactly what appears there. What is missing from your week is missing from the list too – such as how you check results, how you record your ways of working, which rules you set yourself and which content may go into an AI tool. These areas are quieter still than the third pile, because they lack even the task you might notice them by.

For that you need a yardstick from outside. Without something to compare against, any estimate stays a feeling – and decisions about time and money hang on it.

Two questions help with any offer promising an AI maturity assessment – ours included. Does it ask for knowledge or for feeling? A questionnaire that only collects your self-assessment extends the exercise above rather than complementing it. Does it end in a decision or in a report? The only question that counts afterwards is: what do I do first?

Where to begin

Do the exercise today. Ten lines, three marks, one task for next week – that is a complete first step, and it costs you twenty minutes.

For a first look from outside there is the free AI Skill Level Check: five questions, right below this article. If you want a closer look, take the AI Strategy Sonar: around 15 minutes, eight areas, and a share of the questions test knowledge rather than feeling. You receive your profile with strengths and next levers as a document – and a recommendation on what makes sense next and what you can skip. The Sonar is a paid product, one-off per person; the product page gives the price.