The short answer
Do not begin by trying to learn all of AI. Begin with one thing you want to do better: write, research, organise meetings, find a job, run a business or understand the technology. Learn the smallest useful idea, use it on a real task, check the result, and only then decide what to learn next.
That is the free roadmap. It is less impressive than a list of fifty tools, but it is much more likely to survive the second week.
First, choose the kind of AI learner you are
People use “learn AI” to mean at least two different projects.
Using AI means applying existing tools to work and everyday tasks. You learn how to brief a tool, give it context, check its answer and fit it into a routine. You do not need to code for this route.
Building AI means making software that uses models, data and APIs. You should expect Python, data handling, evaluation and a much longer path. It is a worthwhile route, but it is not a prerequisite for using AI well at work.
If your goal is to save time in your current job, do not let a technical course convince you that you need calculus before you can write a useful prompt. You are choosing a different subject, not failing at the first one.
A four-stage free roadmap
Stage one: understand the shape of the tool
Learn what generative AI is good at, what it tends to get wrong and why a fluent answer is not the same as a reliable answer. You need enough vocabulary to make sensible decisions, not enough to reproduce a research paper.
Start with Introduction to Generative AI, then take AI Safety and Verification. Together they answer the two beginner questions that matter: “What can this do?” and “How do I know when not to trust it?”
Stage two: practise giving a clear brief
Prompting is not a collection of magic phrases. It is the skill of explaining a task to someone who has no context: what you want, who it is for, what information matters, what format you need and what to avoid.
Try one task from your actual life. Ask for a first attempt, inspect what is missing, then improve the brief. The improvement is the lesson.
The Prompt Engineering course gives you a short practice route. Do not collect prompt templates just to feel prepared. A good brief you understand is more useful than a folder of prompts you cannot adapt.
Stage three: choose one work outcome
Pick a task you do every week. Examples include turning meeting notes into actions, comparing documents, drafting a customer reply, planning a lesson, preparing for an interview or cleaning a spreadsheet.
Use the tool on a low-risk version first. Keep the original work beside the AI output. Mark what is useful, what is wrong and what still needs your judgement. This is how you learn where the tool belongs in your process.
If you want a guided route, Start Here gives you three courses based on whether you want to save time at work, look for a job, manage a team or run a small business.
Stage four: make the habit safer and more durable
Once a task works, write down the repeatable version: the input, the prompt or instructions, the checks and the final human decision. Never paste confidential information into a tool without understanding the policy that governs it.
Then learn one adjacent skill: verification, editing, spreadsheet thinking, clear writing or process design. Tools change. These skills transfer.
A realistic 30-day schedule
| Week | Focus | Output |
|---|---|---|
| 1 | Understand AI and its limits | One-page list of useful and unsafe tasks |
| 2 | Practise clear briefs | Three before-and-after prompts on real work |
| 3 | Apply AI to one repeatable task | A workflow you can run twice |
| 4 | Check and document it | A short case study: what worked, what failed and what you still review |
You do not need to study for hours every day. Twenty minutes of deliberate practice on a real task will teach you more than passively watching a long playlist.
What to ignore at the beginning
Ignore the pressure to learn every new model. Ignore lists of “the only prompts you will ever need.” Ignore courses that promise a job merely because you completed them. Ignore the idea that a certificate is proof that you can use AI responsibly.
If you later decide to build AI systems, take the technical route deliberately. Learn Python, data and model fundamentals in an order designed for builders. Do not mix that route into your first week simply because it sounds more serious.
Where LearnWithZavi fits
LearnWithZavi is designed for the applied route: short, browser-based lessons for people who want to understand and use AI without a signup or a paywall. The courses are free and the completion certificates are modest records of finishing the lessons, not accredited qualifications.
Start with Introduction to Generative AI, Prompt Engineering and AI Safety and Verification, or let Start Here choose a route based on what you want to accomplish.