How to Learn a New AI Tool Without Losing Three Days to Tutorials
There is a particular kind of procrastination that disguises itself as diligence. You decide to learn Claude, so you watch an introductory video. Then a second video. Then a YouTube channel dedicated to "prompt engineering." Somewhere around hour four, you haven't actually used the tool yet — and you have a meeting in twenty minutes.
This happens constantly with AI tools because the content ecosystem around them is enormous and because learning feels productive even when it isn't. The professionals who actually get good at AI tools do something different.
They use the tool first. They figure it out as they go.
Why Tutorials Waste Your Time
Tutorials are designed for people who don't know what problem they're solving. They show you the full feature set, which is mostly features you will never need, in a sequence that makes sense to the person who built the tool, not to you.
You have a specific job. You have specific recurring tasks that take more time than they should. The most efficient path to using an AI tool well is to point it at one of those tasks immediately and see what happens.
The failure mode is waiting until you feel "ready." That feeling never comes, because readiness is not how adults actually learn skilled tasks.
The 90-Minute Method
When you want to learn a new AI tool, give it ninety minutes of active use on a real task, not a practice task.
Here is what that looks like in practice.
Pick one thing you are already working on today. Something you'd normally spend forty-five minutes to an hour on. A first draft of something, a summary of a document you need to read, a response to a complex email, a research question you've been putting off.
Open the tool. Describe what you need as plainly as you would to a smart colleague who doesn't know your situation. See what comes back. If the output misses, adjust the description and try again. If something confuses you, ask the tool to explain it.
That iteration loop — try, evaluate, adjust — is how you learn faster than any tutorial will teach you. After ninety minutes of this, you know more about how the tool works for your specific work than you would after six hours of watching someone else use it.
What to Pay Attention to In That First Session
You are not trying to produce a masterpiece in the first session. You are trying to learn two things: what does this tool do well, and what does it do badly.
Every tool has both. Claude is strong at nuanced writing and reasoning; it is not a live research tool. Perplexity is strong at finding and citing current information; it is not a writing partner. The faster you understand where a tool earns your trust and where it doesn't, the faster you can use it appropriately.
Make a quick note after your first session. Two columns: what worked, what didn't. This takes five minutes and becomes the mental model you operate from.
One Task at a Time
The second mistake people make — after the tutorial trap — is trying to integrate an AI tool into everything at once. They try to change three or four workflows simultaneously, none of them gets proper attention, and within two weeks they've stopped using the tool entirely.
Pick one task. Make that task work well with the tool before you add a second. For most professionals, this takes two to four weeks of regular use on that one task.
A contracts attorney might start only with drafting first-pass summaries of agreements. A financial planner might start only with drafting client letters. A marketing consultant might start only with building presentation outlines. One task. Repeat it enough times that the tool becomes familiar territory.
When that task feels easy and routine, add one more.
The Prompting Skill Develops by Doing
You will see a lot of content about "prompt engineering" — the art of writing precise instructions to AI tools. Much of it is overthought.
The skill you actually need is describing what you want the way you would describe it to a competent person. You wouldn't hand a colleague a document and say "summarize this." You'd say "summarize the key terms and flag anything unusual for a deal of this type." You would give context. You would specify what matters.
That same instinct — the one you already have from years of briefing people — applies directly to AI prompts. The only learning is translating it. And that translation happens fastest when you are doing real work with real stakes, not practicing with made-up examples.
When You Get Stuck
You will hit moments where the tool produces something unhelpful and you don't know how to redirect it. The fastest fix: tell it what was wrong in plain language.
"This is too formal for the audience." "This misses the main point, which is X." "This is accurate but not the framing I need — I need to lead with Y." You don't need to learn special syntax. You need to describe the gap between what you got and what you wanted.
Most professionals find that after three or four correction cycles on a task, they've learned more about working with that tool than any tutorial would have shown them.
Frequently Asked Questions
How long should it take before I'm actually faster with the tool than without it?
For most professionals using Claude on writing and analysis tasks, this happens within two to three weeks of regular daily use on a single task type. The first few sessions are slower because you're learning. The speed comes after you stop consciously thinking about how to use it.
What if I try a tool and it doesn't produce useful output?
Don't give up after one attempt. Give it three to five genuine tries with different prompts before concluding the tool doesn't work for that task. The output quality is highly sensitive to how the task is described.
Is there anything I should read or watch before starting?
One thing: find out whether the tool has a free tier or requires payment, and what the data privacy terms are for your use case. That's it. Everything else is better learned by doing.
I'm not a fast typist — is that a problem?
Not significantly. The thinking and evaluating take far more time than the typing. If you find typing is genuinely a bottleneck, use voice dictation to write prompts. It works.
What if my organization hasn't approved the tool yet?
Use the free tier for non-sensitive work until you have approval. Once you have a clear sense of what the tool does well, you're in a better position to make the case for organizational access.
If you want a more structured approach to building your AI workflow — including which tasks to prioritize and how to integrate them into your existing workday — the Leverage Starter course ($199) covers exactly this. For professionals who want to build a more complete AI-enabled practice, the Leveraged Associate ($395) goes deeper into applying these tools across client-facing and business development work.
Where this goes next
Want the guided, build-it-this-week version of this? See The Leverage Starter — or Turn Experience Into Income with Claude if you want the broader path.
Related reading from The Briefing
Not sure which program fits where you are? take the 2-minute course-fit quiz, or browse the full TLY course catalog.