50 tools, 10 models, a new promise every week. Anyone who wants to start with AI today drowns first in options. On a phenomenon that has a name - and on the honest first step out of it.
There is a moment we observe in nearly every first conversation. Someone says they have spent the past year intensively engaging with AI - subscribed to newsletters, attended webinars, tested tools. And then the sentence that sums it all up: "I just don't know where to begin."
That is not laziness. It is not a lack of interest. It is a phenomenon in its own right, and it has a name.
Analysis Paralysis - a term from the decision psychology of the 1960s. It describes the state in which someone weighs options for so long that in the end no decision is made. The larger the choice, the higher the felt stakes, the faster the surroundings - the stronger the paralysis. In the AI context this has become a mass phenomenon: the AI paralysis.
Too many tools
ChatGPT, Claude, Gemini, Perplexity, Mistral, Copilot, Notion AI - and 200 more.
Too many promises
Every provider sells its AI as the thing that will transform your business.
Too fast a change
What you learn today may be obsolete in six months. So you wait.
In 2000 the psychologist Sheena Iyengar ran an experiment that every MBA student knows today. In a supermarket, jams were sold - sometimes six kinds to choose from, sometimes twenty-four. The result: the stand with twenty-four kinds drew more people, but at the stand with six kinds people bought ten times as often.
With too many options the brain switches over - from decision mode into comparison mode. And in comparison mode you lose. You start to match every option against every other, to find every weakness, to relativise every strength. In the end you buy nothing, because no option is the best in every respect. With jam that is annoying. With an AI strategy for your business it is expensive - every month in which you decide nothing.
Jam has three properties that AI does not. First: jam does not change on its own. You can compare it for a year, it is still the same. AI tools change quarterly - what you tested six months ago can look completely different today.
Second: jam has clear properties - sweet, sour, fruity. AI tools promise everything. "Boosts your productivity by 40 percent." No one can prove that beforehand. No one can measure it afterwards. You have to believe it or try it out, and in trying it out you are already caught in the next comparison mode.
Third: jam costs 4 euros. Introducing AI costs time, licence fees, training, the reworking of processes, perhaps changes to staffing structures. The stakes are high, and the higher the stakes, the stronger the fear of overreaching. The result: you wait until "it becomes clearer". But with AI it does not become clearer. It becomes faster.
AI paralysis shows itself in a recurring pattern. The most common symptoms are:
Tab-hoarding. You have 30 browser tabs open with AI articles you mean to "read later". You never read them. Instead, new ones are added every day.
Newsletter bingo. You subscribe to five AI newsletters, because each gives you the feeling you will miss something if you do not have it. You read none in full.
Tool-hopping. You test ChatGPT one week, Claude one week, Perplexity one week. Each time you reach a point where you think "maybe the next one is better" - and you switch before you have really used the current one.
Waiting for "the right moment". You tell yourself: when GPT-5 is out, I'll begin. When the EU AI Act rules are clearer, I'll begin. When my tax adviser knows their way around, I'll begin. The right moment never comes - because the field never stands still.
Webinar mania. You book an online event every week, hoping to find clarity there. Instead, each time you get the feeling of falling even further behind.
Behind AI paralysis there is often FOMO - the fear of missing out. Every tech influencer on LinkedIn writes that whoever does not get in now will be left behind within a year. Every headline suggests that your competitors are long since ahead. You see examples of solo entrepreneurs earning six figures with AI - while you are still sorting your tabs.
The truth is more sobering. Studies show: most businesses that invested early in AI lost money doing so - because they started without a clear application. "We're now doing something with AI" is not a strategy. It is a reassurance against the fear of doing nothing. And it usually costs more than it brings in.
The businesses that draw real value from AI today are not the fastest. They are the ones that had a concrete question before they began with AI. "Where do I lose the most time?" is a better starting question than "Where could I use AI?"
Anyone who wants out of the paralysis needs no better market overview. They need a different order.
First move: begin with a problem, not with a tool. Write down where this week you lost the most time. Which task annoyed you most. Which request you have answered five times because it keeps coming back the same. That is your AI task. Not what the newsletter says.
Second move: choose one, not the best. There are today three or four serious all-round AIs (ChatGPT, Claude, Gemini). They can do the same in 80 percent of small-business tasks. Which one you pick matters far less than that you pick one and start with it. If after three months you notice the other one fits better, you switch - with experience behind you. But first: do it.
Third move: stay small, stick with it long. No one becomes an AI pro in two weeks. But anyone can, in two weeks, automate one concrete task. When that runs, the second comes. When the second runs, the third. Depth beats breadth - in a field where there is endless breadth and very little depth.
Just as important as the first step is the permission to let some things go.
You don't have to know all the models. You don't have to know whether GPT-5 or Claude 4 is better. That is the job of AI researchers. Your job is to use one of them.
You don't have to try every new tool. Ten new ones come every week. 90 percent of them will no longer exist in two years. If something were really relevant to your business, it would meet you after three months from three different directions - then it is time to take a look. Not before.
You don't have to do everything yourself. It is no weakness to let yourself be accompanied. No one expects a building contractor to be structural engineer, architect and plumber in one person. With AI it is the same. Anyone who wants to get ahead brings in someone who has the overview - so that you don't have to be one of those who monitor the field every day.
Our whole approach is an answer to AI paralysis, even though we did not call it that. Three points in concrete terms:
First the tools, then the AI. We never build AI on a shaky foundation. First we clear up the processes with you - what runs, what jams, what is even sensibly automatable. Then, and only then, AI comes into play. Once the base stands, the tool decision is, in 90 percent of cases, easy - because the solution almost arises from the problem on its own.
We decide for you where you want us to. We take the market-watching off your hands. You don't have to know whether a particular model has just got better - that is our job. You tell us what you need. We tell you with which tool that works best today. That is not paternalism - that is division of labour.
We stick with it. A one-off installation does not solve AI paralysis. When the field changes further, your tools change too - but not every week, rather when a switch is really worth it. We watch, you work. That is how it works.
The question is not which tool is the best. The question is whether in twelve months you will be using one - or still comparing.
AI paralysis is no weakness. It is a normal reaction to an abnormal market situation. Whoever overcomes it gains not through speed, but through clarity. One problem, one tool, one first step - that is more than 90 percent of all those who have, for a year, wanted to do "something with AI" will ever achieve.
If you want to get your bearings first, read on wendwerk.de/wissen the texts Where to begin with AI and What AI makes possible. If you want to know which voices are shaping the AI world right now, read Where we are headed.
Curated by Johannes Hohls for wendwerk.