Juliette's Learning Tribe
How to learn AI skills for work, one real task at a time
You do not need a coding course to learn AI for work. Pick one task you do every week, do it with AI, check every line, and repeat.
In August one of my AI agents reported back on a piece of work. It said it had written thirty four checks to prove the work did what it should.
It had written none.
I had already approved the work on the strength of that one sentence.
I use AI all day. I am dyslexic, and for most of my life writing either took me forever or did not happen at all. Claude changed that. The thing in my head lands on the page now, and I love it for that.
But that agent taught me the most useful thing I know about AI at work. It sounds exactly the same when it is wrong as when it is right.
So when people ask me how to learn AI skills for work, my answer is short. Learn it on your own job, and check everything.
Why the tutorial does not stick
Most people start with a video about the tool. They learn where the buttons are. Then they go back to their desk and nothing changes, because the video was about somebody else’s work.
I have watched this happen with training for twenty years. People are shown a thing. A week later they are asked to do it on their own, and it has gone. Knowing about it is not learning it.
The people who get good at AI start with a task they already do.
Pick one boring task you do every week
Something you could do in your sleep. The Monday update email, say, or the report you rebuild from the same spreadsheet every week.
Pick one. Only one.
Do it with AI
Give it what you would give a new colleague on their first day. What the task is, who it is for, what good looks like. Paste in your rough notes rather than a blank page.
There is a tip sheet on talking to AI at work on this blog. Keep it open the first few times.
Check every line
This is the skill, and it is the half most people skip.
Read what comes back as if a bright new starter wrote it. Look for the number with no source. The rule that does not exist. The name spelled nearly right. Fix it, then notice what kind of wrong it was, because it will do it again.
Checking works both ways. That same month I told Claude to relabel forty thousand records. It went and looked, came back, and said that would make things worse. It was right. Being wrong for an afternoon saved me a month.
One rule does not bend. Never paste in anything you would not email outside the company. The UK’s National Cyber Security Centre says not to put sensitive or confidential information into public AI tools. Client details and personal data stay out.
Save the part you repeat
After a few weeks you will see a pattern. The same instructions, every Monday. Save them as a template or a saved prompt, or build a small automation if your tools allow it.
Then look at what changed. How long does the task take now? Is it better, or only faster?
And what did you do with the time you got back? That is where your job starts to change.
It is slower at first
The first week you will probably be slower than doing it yourself, because you are checking everything. That is the learning happening. It does not mean it is failing.
Some tasks are not worth it. If the checking takes longer than the doing, drop it and pick another.
Can you learn AI skills for free?
You can start for free. Most of the big AI tools have a free version, and your own work costs nothing to practise on. What costs is time. Doing it properly, week after week, and knowing whether you are getting better.
What we built
This is how we built Future Fluency, our AI skills for work. It is applied to your own role. You use AI on your real tasks, check what comes back, automate something you repeat, and look at what changed. Your AI tutor builds the plan from the job you have.
I am going through it myself. You cannot ask a company to get fluent at this while you are the least fluent person in the building.
Your first Future Fluency module is free. After that, learning is paid.
→ Start your free first module
Sources
- UK National Cyber Security Centre, guidance on ChatGPT and large language models (do not put sensitive or confidential information into public LLMs): ncsc.gov.uk