If you can write a clear email and follow written instructions, you can learn to build with AI. There is no maths gate, and the tools are in English (and increasingly every major language).
What this roadmap will not do:
What it can realistically do in 3 months of consistent effort (about 30-45 minutes a day):
If you were hoping for a shortcut, this is not one. If you were hoping for a real path someone without a technical background can follow, it is.
Time-to-first-payment ranges by platform, from the sources cited in this article.
Read this carefully, because most roadmaps skip it.
Most people who try earn little or nothing in the first month. That is normal, not failure.
No sign-up, nothing to install, and it takes about a minute. This is the same thing the course has you do on day one.
You'll end up with: five quick tests that show what this AI is actually good and bad at, and what to watch for in each answer.
Runs on bloom0's own AI. Nothing you type is stored or shared.
Ignore anything that does not build one of these:
1. Prompting well. Writing clear instructions to a model — including what format you want back, what to avoid, and what context matters. This is 80% of practical AI work.
2. Chaining steps. Using AI for one part of a task (e.g. summarise), then another (e.g. rewrite in a client's voice), then a third (e.g. format for their platform). Real client work is almost always a small chain, not one prompt.
3. Checking output. Spotting when the model is wrong, vague, or made something up. This is exactly what DataAnnotation pays for and what every client silently needs.
4. Packaging. Turning a workflow into something someone would pay for: a Fiverr gig, an Upwork proposal, a one-page service description.
You do not need to learn Python. You do not need to learn how transformers work. You need to get very good at these four.
Weeks 1-2 — Learn the tools (about 30 min/day). Pick one chat model (ChatGPT, Claude, or Gemini) and use it daily for real tasks in your life: emails, planning, summarising articles. Learn what it does well and where it fails.
Weeks 3-6 — Build 3 small projects. Something like: a job-application email generator for a friend, a spreadsheet cleaner, a study-notes summariser for a student, a product-description writer for a small shop. Each one should solve a specific problem for a specific person. Save your prompts and the results.
Weeks 7-9 — Package one project. Turn your strongest project into a narrow offer. Not "AI services" — something like "I turn 60-minute meeting recordings into 5-bullet summaries for £25".
Weeks 10-12 — Go where the money is. List the gig on Fiverr, send 3-5 targeted Upwork proposals a day, and in parallel apply to DataAnnotation. If you have professional expertise, apply to Mercor too.
If you want a structured beginning that is not another endless video course, bloom0 is a free bilingual course that walks complete beginners through building real things with AI. It is designed for the weeks 1-6 part of this roadmap — the part where most people give up because they do not know what to build.
Whatever you use, the rule is the same: build small things for real people you know, before you try to sell to strangers. A cousin who runs a small business, a friend applying for jobs, a parent drowning in admin — these are your first "clients". They give you honest feedback, and honest feedback is the difference between a portfolio that wins work and one that does not.
Total time: about 460 minutes.
Three questions. No score is stored, nothing is sent anywhere.
1. You have 30 minutes today to move forward on this roadmap. What is the best use of it?
Building — even tiny, imperfect building — teaches you more than any amount of watching or comparing. The prompts you save today become the raw material for a portfolio project later. Comparison articles feel productive but rarely change what you actually do.
2. Someone with no professional background wants the highest hourly rate as fast as possible. Which platform is the wrong first target?
Mercor's high rates go to verified professionals in fields like medicine, law and finance. Without existing domain expertise, it is a month-3+ target, not a starting point. Fiverr and DataAnnotation are more realistic first stops for a true beginner.
3. You have built three small AI projects. What is the single most important next step?
More projects without a paid offer is a common trap — it feels like progress but delays income indefinitely. The leap that matters is packaging what you already have into a specific offer someone can buy. Code can come later, only if a real client actually needs it.
Free forever · no account needed · nothing you make is shared