7 min read · updated 2026-08-17 · free, no sign-up
Short answerFiverr CEO Micha Kaufman's 2025 internal memo said AI is coming for easy tasks in every profession, including his own, and told staff to become 'exceptional' at using it. For someone learning AI now, the practical read is this: the low end of freelance work is being squeezed, but people who can actually operate AI tools to deliver a result are still in demand — and that skill takes weeks, not years, to reach a useful level.
What Kaufman actually said, in plain English
The memo was blunt. AI will do the easy parts of most jobs, so the safe position is being one of the people who uses it well. He wasn't predicting the future — he was telling his own employees to adapt now or fall behind.
For you, that reframes the question. The query isn't really 'should I fear AI', it's 'what side of this line do I want to be on?' The line runs between:
People who only do the easy task (writing a basic bio, editing a simple image, translating a paragraph). Their rates are getting compressed.
People who use AI to deliver something a buyer couldn't do themselves — a working automation, a full content system, a cleaned dataset, a fixed website.
The good news for a beginner: you don't need a computer science degree to cross that line. You need a narrow skill, a clear offer, and a few weeks of unglamorous practice.
Where beginners can realistically start
Reported pay ranges and entry barriers for four AI-adjacent platforms, from the sources listed at the end.
Is this real income or another hype cycle?
Both, honestly. There is real money moving through AI-adjacent freelance work right now, and there is a lot of noise around it. Here's what the verified numbers actually say:
Upwork's own rate guidance: entry-level freelance work sits at $10–$25/hr, intermediate at $25–$75/hr, and specialised AI, development and consulting work often reaches $75–$150+/hr (Upwork).
Mercor, a marketplace matching experts to AI labs, reports rates from ~$25/hr up to $200/hr — but the top end is for existing professional expertise like medicine or law (review).
DataAnnotation, which pays people to correct AI outputs, commonly reports $20/hr base (review), with 1,500+ Indeed reviews as proof it pays real people.
The honest part: most beginners earn little or nothing in their first weeks. The people who quit are usually the ones who expected week-one money.
Make one right now, on this page
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: A proposal you can send to charge more.
Runs on bloom0's own AI. Nothing you type is stored or shared.
Where beginners actually start (and where they shouldn't)
Four real doors, ranked by how fast a complete beginner can walk through them:
DataAnnotation — Easiest entry. No degree, but you must pass an unpaid assessment, and that's where most people fail. Work availability is inconsistent — treat it as variable income, not a salary.
Fiverr — No gate to list a gig. The barrier is being invisible until you have your first few orders. Expect to price low at first to earn ranking.
Upwork — Also no gate, but you're bidding against thousands. Winning a first contract with zero reviews is the real hurdle. Narrow offers win.
Mercor — Best rates, but genuinely aimed at people who already have a profession. If you're brand new, this is a month-3 goal after you've built something to show.
What to skip: any 'AI course' that promises a specific income figure. None of the sources above do that, and neither should anyone selling you a shortcut.
The skill that turns AI use into paid work
Kaufman's memo implies something important: prompting is not the skill. Anyone can prompt. The skill is packaging AI output into something a buyer will pay for and not have to fix.
Concretely, that means learning to:
Chain steps. Real work is rarely 'one prompt in, one answer out'. It's research → draft → check → format → deliver.
Verify. AI is confidently wrong often. Buyers pay you partly to catch that.
Package. A Google Doc with tracked changes looks professional. A raw ChatGPT paste does not.
Narrow the offer. 'I write with AI' loses to 'I write 800-word LinkedIn posts for B2B founders, three per week, in your voice'.
This is what bloom0's free bilingual course is built around — moving beginners from 'I've used ChatGPT' to 'I can deliver a real thing'. The gap between those two takes most people a few weeks of focused practice, not months.
A realistic first month
If you do the following steps in order, you'll know within 30 days whether this path is for you — which is more than most people can say after a year of watching YouTube.
The plan below assumes about 30–90 minutes a day, most days. If you have less, stretch it over six weeks instead of four. Don't try to compress it; the assessments and applications have wait times you can't control.
One warning: do not open five platforms at once. Beginners who apply everywhere end up with five half-finished profiles and nothing to show. Pick the door that fits your situation and get through it first.
What to actually do
Total time: about 235 minutes.
Day 1Pick one door. If you want the fastest test of 'can I earn from AI at all', start DataAnnotation's assessment (site). If you'd rather build a sellable skill, skip to Day 2. Do not sign up for more than one platform this week.40 min
Days 2–7Choose one narrow offer you could deliver with AI help — e.g. 'clean up messy spreadsheets', 'summarise long PDFs into one-page briefs', 'write LinkedIn posts for accountants'. Deliver it three times for free to friends or small businesses. Keep the outputs — these become your portfolio.60 min
Week 2Turn your three free deliveries into one Fiverr gig or one Upwork profile. Price low on purpose. Write the offer in the buyer's language, not yours ('I turn your 40-page report into a 1-page brief in 24 hours'), not 'I use GPT-4'.90 min
Weeks 3–4Send 3–5 targeted proposals per weekday on Upwork, or refresh and refine your Fiverr gig based on views vs orders. Track what you send and what replies. If nothing lands after 30 real proposals, the offer is too broad — narrow it further, don't send more.45 min
Check yourself
Three questions. No score is stored, nothing is sent anywhere.
1. A friend says 'I'll list an AI writing gig on Fiverr tonight and start earning tomorrow.' What's the most useful thing to tell them?
Fiverr has no barrier to listing, which is why people expect fast results. The real barrier is discoverability. Independent seller reviews show meaningful revenue only tends to appear after several completed orders build ranking — early low prices are a deliberate ranking strategy, not a failure.
2. Which of these is closest to what Fiverr's CEO actually urged in his 2025 AI memo?
The memo's core message was about adaptation, not replacement or coding. This matters for beginners because it points to the actual opportunity: not building AI, but using it well enough to deliver results buyers will pay for.
3. You have no professional background yet but want the highest-paying AI-adjacent work. Which is the honest first move?
Mercor's top rates ($200/hr) go to verified professional experts in fields like medicine and law. Applying without that background wastes an application. Building a narrow, evidenced skill first on a lower-barrier platform is how beginners actually reach specialist rates over time.
Free forever · no account needed · nothing you make is shared
Money figures on this page come from the sources listed below and are ranges, not promises. Most people earn little or nothing at first, and platform rates change. Always check the source before you count on a number.