"It’s too complicated": why your first step should be smaller than you might imagine

znn-os · JARVIS — Tue, 27 Oct 2026 09:21:51 +0000

Most people who claim they don’t have time to look into AI are neither lazy nor behind the times. They are simply standing before a hundred different doors, with no indication of what lies behind them. Here is how to choose one.

What lies behind these words

When professionals say that "AI is too complicated," they don’t necessarily mean that the technology itself is difficult to use. Entering a prompt into a chat interface is not complex at all. What they are actually expressing is more like this: there is too much to learn, and I can’t figure out what is truly relevant to my day-to-day life.

This is a different kind of problem, one that requires a different solution. If the difficulty is the issue, then things need to be simplified. But if the problem is, "I don’t know where to start, and I can’t afford to waste an afternoon searching," then what you need isn't necessarily a better tool. You need a more modest first step.

This article explains how to identify that step—and how to proceed once you’ve found it.

First step: choose a task, not a tool

The most common approach is to start by choosing a tool. You compare several options, read reviews, watch tutorials... Weeks go by, yet nothing has actually changed in your daily routine.

Flip that logic around. Look at your calendar and ask yourself three questions about the past seven days:

Which task did I perform more than once?
What task have I put off—not because it was difficult, but because it was tedious?
What task involves taking existing textual content (an email, a note, a list) and turning it into a text intended for someone else to read?

The third question is the most important. Current AI assistants are particularly effective at processing textual information: they take in text and produce text in return. A task that follows this pattern is an excellent starting point. Conversely, a task that relies on your own unique judgment—or on information that hasn’t been recorded anywhere—is not a good fit.

State this task in a single sentence. If the sentence requires the word "and," it actually encompasses two separate tasks. Choose just one.

An example: the follow-up message

The scenario below is an illustration designed to demonstrate the method. It is not a real-life account, nor does it represent an actual outcome.

Consider a freelance translator. After every delivery, she is supposed to send a brief message to the client specifying what was delivered, noting any assumptions made regarding ambiguous terms, and including a polite closing statement indicating that she remains available for any questions. She knows she should do this every time, but—often feeling tired—she ends up forgetting.

Let’s apply the three questions. This task recurs regularly. She tends to put it off because she finds it tedious. The process begins with information she already has (her own notes on the work performed) and results in a message intended for the client. This fits the method perfectly. The first step is not to "integrate AI into one's professional practice," but rather to do this: copy and paste notes from a completed assignment and ask for an initial draft of a follow-up message.

Next comes a step that is all too easily overlooked: reviewing the draft with a critical eye—acting more like an editor than a mere user. Does the text contain points phrased in a way you wouldn't have chosen? Did it invent any details? Make the necessary corrections, send the message, and then jot down a single line noting the changes you had to make. This line is the most valuable part of the entire process, as it tells you exactly what information to include in your prompt next time.

None of this requires acquiring skills in a new field. All it takes is a single task, some input data, and a careful review.

Step Two: Add Context, Not Complexity

The second time around, she doesn’t switch tools. Instead, she refines her prompt by incorporating the element she previously had to correct: "Limit the response to four sentences. Do not mention deadlines unless I specify one. Use a tone that is warm yet formal."

This is how you make progress. With each iteration, the prompt becomes slightly more precise—not because a guide advised you to use a specific technique, but because you discovered, through experience, what was actually necessary. After a few attempts, you end up with a prompt that works for you, phrased in your own words and tested against your actual work.

This represents a level of mastery far beyond what most people imagine—and you’ve achieved it without needing to learn any technical jargon.

Step Three: Consider Taking It Further

Once a task is well-mastered, a logical question arises: what is the next task that resembles this one? For example: drafting a weekly summary of your notes, formulating an initial response to a recurring request, or creating a checklist based on a previous project.

Notice how this "mastery" was built: it wasn’t the starting point, but rather the result of accumulating small, well-executed tasks. Down the line, you might want to entrust the assistant with longer sequences of steps, or have it interact directly with files and applications. This is a perfectly natural evolution—and it’s much easier to evaluate once you understand where the assistant is reliable and where it requires human verification. If you choose this path, keep in mind the rule that made the first step safe: always verify the result before passing it on.

Three important limitations to keep in mind

Verify before sending. An assistant may generate perfectly fluent sentences that are nonetheless factually incorrect. Fluency is not proof of accuracy.
Be cautious with the data you input. If your work involves confidential client information, consult the tool’s terms of use and any professional guidelines that apply to you before entering such data. When in doubt, start with tasks that do not contain sensitive information—such as working with templates, public texts, or your own personal notes.
A modest first step does not guarantee an ideal outcome. Some tasks are simply not worth automating. Discovering this within twenty minutes—while working on a tedious task—is, in itself, a very positive result.

What to do this week?

Make a list of the tasks you completed last week and identify those that involve producing or processing text.
Choose the most tedious one. Summarize it in a single sentence.
Perform this task once using real data, examine the result with a critical eye (as an editor would), and note the changes you made.
Perform it a second time, incorporating those changes directly into your prompt.

Four steps, one single task, no new jargon. If, after this, the process still feels too complex, you will at least know which specific part is causing the issue—which is far better than simply concluding that "everything is complicated."

This article was written for the JARVIS project, a voice and visual interface designed for macOS:
https://znn-os-jarvis.norbert.md

This article was written with the help of an AI model and reviewed by its author before publication.

DEV Community: znn-os · JARVIS