Most people were handed the tools and no instructions. The useful split is between learning the tool and learning to judge what it gives you, because only one of those transfers.
Want to get better at something specific?
Try Cadence for a course built around your goal, with short lessons, practice, and saved progress. Start free on iPhone or request Android access.
Learning Cadence editorial, , 9 min readHow we work
AI training at work tends to arrive as a tool demo. You learn where the buttons are, and you still cannot tell whether the output in front of you is any good. That second thing is the skill that matters, and it is specific to your own work.
So the options below split by what you are actually trying to build: judgement about output in your field, fluency with a particular tool, the ability to write code alongside a model, or a credential someone has asked for. They are different jobs and the right pick is different for each.
WhyFree sequences and practice across the underlying subjects.
The apps
Every app, with what it is bad at.
Start with the job you need the app to do, then read the tradeoff.
01
Cadence
A course built around your starting point
This is our app. See the disclosure above.
Cadence starts with a problem you want to solve or a curiosity you want to explore, then builds an ordered 7 to 30 lesson course around it. Each lesson takes about ten minutes, can be read or played as audio, and stays connected to the course. Lessons use checks or practice when they help, and Ask Coach stays inside the lesson.
Best for
Turning your own starting point into one coherent course you can keep progressing through.
Not for
Anyone who needs a recognised certificate, a named instructor, or a language course with speaking practice.
Price
Free to start, with unlimited daily lessons and audio and 4 lesson voices included, and 2 new custom courses a month. Plus adds 8 more voices, for 12 in total, and makes new courses, practice, and Ask Coach unlimited for US$49.99 a year on the web, with a 7-day trial.
Mimo teaches Python, JavaScript, HTML, CSS, SQL, React, Swift, and related development skills through guided paths, short coding exercises, and projects. You write code during the lesson, then combine the ideas in work you build yourself.
Best for
Starting coding through small interactive exercises before moving into larger projects.
Not for
Learning software development without eventually building and debugging projects outside the guided lesson.
Price
Free to start. Paid plans add broader access to paths, projects, certificates, and career-focused features.
Coursera hosts courses from universities and companies, with video lectures, readings, assessments, and paid certificate tracks. The first module of nearly every course can be previewed free, which makes it useful for checking the instructor and workload before paying.
Best for
Subjects where the name on the certificate matters to an employer or an application.
Not for
A ten-minute daily habit. Most courses expect real study blocks.
Price
The first module of nearly every course is free to preview. Continued access, graded work beyond the preview, certificates, and subscription plans are generally paid.
Practical and technical skills, bought one course at a time
Udemy is a marketplace, so anyone can publish a course. Quality varies a lot between instructors, which makes the ratings and the free preview lessons worth using before you buy. It is strong for software, tools, and other subjects where you want a specific technique.
Best for
One concrete tool or technique you need soon.
Not for
Consistent quality without doing your own checking, or a daily learning habit.
Price
Courses are bought individually and are discounted often.
Brilliant teaches through problems and not lectures. You are given a puzzle, you try it, and the explanation arrives once you have already thought about it. That order is harder work than watching a video and it is why the ideas tend to stay.
Best for
People who want to think during a lesson instead of watching one.
Not for
Subjects outside maths, science, and computing, or learners who need unlimited daily access without a subscription.
Price
Brilliant Free provides two keys a day. Each key unlocks one lesson or practice set, and free users move through courses in order. Premium removes the daily limit, allows jumping ahead, removes ads, and gives full Koji access.
edX was founded by MIT and Harvard and works much like Coursera. The audit track lets you follow a university course for free for a set period. Verified certificates and full programmes are paid.
Best for
Following a real university syllabus without paying for the badge.
Not for
Anyone who wants permanent access, since audit access is usually time limited.
Price
Auditing is free on most courses. Certificates and programmes are paid.
Khan Academy is a nonprofit that publishes full course sequences in maths, science, economics, and test preparation. Every lesson is free, there are no ads, and the practice sets tell you which idea you have not understood yet instead of only marking you wrong.
Best for
Rebuilding a foundation you missed at school, especially in maths and science.
Not for
Soft skills, workplace situations, and creative subjects.
Price
Free. It is funded by donations, not subscriptions.
Start from a task you already repeat, not from the tool.
A course about AI in general teaches the tool. Your difficulty is where the tool meets your own job, and that is where all the useful judgement lives. It is why people finish an AI course and still cannot say whether the draft in front of them is good.
Pick one thing you do every week. Write the instruction you would give a new colleague: what the output is for, who reads it, what a good one contains, what to leave out. Then run it on a case where you already know the right answer, because that is the only way to find out whether the output is right rather than merely plausible.
Does the lesson leave you with the actual sentence you would use, or a summary of an idea?
Can you give the app your own situation rather than a generic example?
Is there practice, or only a quiz that checks whether you read it?
Do you get a response to your attempt, not just a model answer?
A caution
Judging output is hard exactly when you are new.
Everything above depends on having ground truth. That is straightforward for work you have done for years and genuinely hard when you are new, which is when the tool is most tempting and most risky.
There is no clever way around it. If you cannot yet tell a good answer from a plausible one in your field, using AI to produce that answer will not teach you, and may cost you the judgement you were about to build. Learn enough of the underlying skill to have an opinion first, then use the tool to go faster.
Where we fit
Try a Cadence course on your own AI task.
We make Cadence. It builds a course around a goal you describe, so the examples can be about the weekly update you write or the replies you draft, rather than a generic prompt library. Each lesson carries practice and a coach you can ask about your own case.
It does not replace the checking work above, and nothing does. If you need a named instructor, a specific tool demonstration or a qualification an employer requires, check a specialist course first.
Put it into practice
Start it today, ten minutes at a time.
Cadence builds a course around your own goal instead of a catalogue, with short daily lessons, practice and saved progress. Cadence is our app.
Questions
Common questions.
What is the best app to learn AI skills for work?
It depends what you are building. For judgement about output in your own field, a course built on a task you already repeat beats a general AI course. For coding alongside a model, Mimo teaches through exercises. For a credential, Coursera and edX carry university and company courses, but ask which one your employer accepts first. For one specific tool, a single Udemy course is usually faster than a subscription.
What AI skills do employers actually want?
Less prompt trickery than judgement: being able to say what a good output for your job contains, spot where a confident answer is unsupported, know which decisions still need a person, and work the tool into a task without losing the ability to check it. Those transfer across tools. A memorised prompt does not.
Do I need to learn prompt engineering?
You need to write clear instructions, which is mostly clear thinking about what you want. Stating the purpose, the reader, what a good result contains and what to leave out matters far more than elaborate technique. Borrowed prompt libraries can produce a decent result without teaching you why, which leaves you stuck when the task changes.
How do I know if an AI answer is wrong?
Test it where you have ground truth. Ask for the reasoning and check each step follows. Ask what would have to be true for the answer to be wrong. Verify one checkable fact before trusting the parts you cannot check. If you cannot yet judge the answer in your field, learn enough of the underlying skill to have an opinion first.
How long does it take to get good at using AI at work?
It depends how often the task repeats, because the learning comes from the comparisons rather than the hours. A task you do weekly gives you a useful read within a few cycles. There is no credential at the end of this, and anyone promising a fixed timeline to competence is guessing.
See it in concrete terms
Open a real Cadence course.
These Cadence courses cover AI at work and the data judgement underneath it, one lesson at a time.
Options are grouped by what you are building: judgement in your own domain, tool fluency, coding alongside a model, or a credential. This is an editorial decision aid, not a claim that we measured learning outcomes.
What we checked directly: Each app is described from its current official pages. We did not test every course inside these catalogues.
Limits of this review: We did not buy every subscription, and we did not evaluate individual instructors. No exact competitor prices are quoted because they change by region, platform and offer.
