Three tracks: which direction to choose if you can do a bit of everything
A person completes the training, can make images, clips, texts and a few bots — and gets stuck, because he does not understand which of it to sell. Below are three tracks along which the work comes, with prices, time to first money and what is required to start. Plus a rule for choosing in ten minutes.
A person completes the training, can make images, clips, texts and a few bots — and gets stuck, because he does not understand which of it to sell. Below are three tracks along which the work comes, with prices, time to first money and what is required to start. Plus a rule for choosing in ten minutes.
"I finished the training and do not know which field to settle on"
That is a direct quote from my survey. Two more sit alongside it: "unclear how and where to move" and "no system for earning".
Note what is absent from those phrasings. There is no "I cannot do it". The person can. He cannot choose — and that is a different problem, one training does not cure.
The choice is hard for a clear reason. Tracks get compared by interest, when they should be compared on three things at once: how soon the first money arrives, what the fee ceiling of the direction is, and what is required to start. Those three differ across the tracks, and a different one wins on each.
Below are all three, from a scrape of 5 986 jobs over four months.
Track 1. Video and UGC — the fastest entry
What it is. Short clips for social media, advertising from a product photo, editing other people's material, voiceovers, a talking character.
Why it is the fastest. Video is the most common work in our bot's feed: the "AI creators" category, which includes video, is 86% of jobs, editing in another 16%. Demand is daily, the task is small, and the solution is verified in a minute of watching. The client does not need to believe you — he looks at the result.
The money.
| Work | Range |
|---|---|
| Editing supplied material | $12–35 |
| A short clip | $18–45 |
| A 2–3 minute animation | $25–110 |
| Turnkey AI video | $245–550 |
What is needed to start. Three to five clips of your own. Not for a client — for yourself, simply so there is something to show.
The downside to know in advance. The bottom rows of that table are the cheapest on the market, and competition there is highest. The track gives first money quickly and second money slowly: to reach $245–550 you have to stop selling a clip and start selling a series.
Track 2. Sites and vibe coding — the highest fee for one-off work
What it is. A landing page, a multi-page site, a portfolio site, a simple interface, an application prototype.
Why people come here. One-off work costs several times more: $245–730 for a multi-page site against a median one-off job of around $60. One deal covers a month of applying for clip work.
What is needed to start. Two assembled sites, preferably not from templates. That takes longer than three clips — count on a couple of evenings each.
The downside. The deal cycle is longer. A site is not ordered "for today", decisions take weeks, and revisions arrive in waves. You have to be ready for that — the money here is larger but it arrives later.
Who it suits. People not desperate to close their first hundred dollars this week, and comfortable with a client vanishing for ten days.
Track 3. Automations and assistants — the highest ceiling
What it is. A chain of services doing the work without a person: collecting enquiries, analysing calls, reports, a bot on the company's knowledge base.
Why the ceiling is higher. A system or a pipeline is $490–1 460. And it is the only track where the work does not end at handover: what is built has to be maintained, and maintenance is payment every month rather than a one-off fee.
What is needed to start. One assembled, working automation. Your own, on your own tasks — that is enough to show a client the logic.
The downside, and it is a serious one. Selling here is hardest. The client does not see the result with his eyes as he does a clip — it has to be explained to him what exactly he is buying. This track is not for someone who wants first money this month.
The three tracks side by side
| Video and UGC | Sites and code | Automations | |
|---|---|---|---|
| First money | fastest | medium | slowest |
| One-off fee | $12–550 | $245–730 | $490–1 460 |
| Repeatability | a new job each time | a new job each time | a retainer |
| Needed to start | 3–5 clips | 2 sites | 1 automation |
| Hardest part | competition | long cycle | explaining what is bought |
How to read the table. No row makes a track better. If money is needed this month, the first column. If you can wait a month or two for a larger fee, the second. If you have already automated something and enjoyed it, the third.
The ten-minute rule for choosing
Do not guess which has better prospects. Ask yourself two questions.
First: which of these three have you already done for yourself, successfully? Not in theory — with your hands. Usually one direction comes out. That is the one.
Second: how many hours a week do you have? By the survey, 61.3% fit within five hours. Five hours is the first track. Automations can be assembled on that schedule, but slowly, and the first job will not come soon.
If both answers agree, the question is closed. If they diverge, the first wins: selling what you have already done is safer than selling what is more interesting.
What clients actually require
A useful figure that removes half the doubt. A portfolio is asked for in 37.5% of listings, experience in 3%. Roughly twelve to one.
The client does not care how many years you have been in the profession or how many tracks you tried. He cares about seeing that similar work has come out well for you before.
What to do if you chose wrongly
This is the most common fear, and it is overstated.
The niche changes after the first jobs, and that is normal: choosing badly and correcting is faster than not choosing. Three months without a choice cost more than a month in the wrong direction, because a month in a direction produces data — applications, refusals, clients' questions, real timescales. Three months of doubt produce nothing.
And the tracks overlap more than they appear to. Someone who built sites sells automations to the same clients. Someone who made clips arrives at series and positioning. The first track is not a lifelong choice but an entrance.
What to do today to pick a track
Step one — choose one track by the two questions above. One, not two.
Step two — assemble the entry. Three clips, two sites or one automation. For yourself, on your own tasks, in an evening or two.
Step three — ten applications in the category you chose. Without picking the ideal job, without preparing for a week.
If you cannot choose. Two assistants solve exactly that, Basic plan. AI world navigator looks at what you already know, rates your level with each tool and shows which combinations add up to a marketable skill. Online income mentor takes it further: what interests you, what you are good at, which track suits that.
An analysis of the ten hottest niches with their fee levels — the course “Sales”, the lesson “Clients and the market: the top 10 niches and the secrets of communication”, Basic plan. Worth looking at before choosing: the tracks differ not only in what you do but in what it pays.
The express course “Freelancer positioning” ($9, Basic plan and above) starts from the same point: the Personality unpacker and the Skill map define the niche on the "AI skill × Domain × Format" formula before you begin doing anything.
Tools for each track sit in one place. Assistants for the task, ready templates, the "Freelancer positioning" express course for the entry and the ten-lesson sales module for the first negotiations.
And what does not depend on the track chosen — the practical deal breakdown — the “First order” series in the course “Sales”, six lessons: the path from finding a first client to presenting the finished work and an upsell, walked in full on screen. The track determines what you sell; that breakdown is how you sell it, and it is the same for all three.
Registration is free and opens three days of full Basic access. Three days is enough to assemble the entry for your chosen track and send your first applications.
Choose one track today. Not the best — yours. The market will show you the rest faster than any amount of reflection on prospects.
What to take on the platform for the track you choose
Video and UGC
- Course “Photo and Video” — Plus plan: from the basics of generation to AI video and the video pipeline
- Course “Video editing” — Plus plan: trimming, cutting, subtitles, sound, colour correction
- Express course “Google Veo 3 — creating commercial video” — $9, Plus plan and above
Sites and vibe coding
- Course “VIBE CODING” — Creator plan: from an idea and a technical brief to building and accepting a first app
- The “Site templates” section — 343 prompts with previews, Plus plan
Automations and assistants (regular templates come with the Creator plan, and on Make, N8N and Full)
- Course “Automations in n8n” — N8N plan; course “Automations in Make” — Make plan (Full opens both)
- AI routing of incoming requests by department — an n8n template: sorting and distributing requests
- Vapi — saving a call summary to a spreadsheet — a Make template: call results into a sheet
- Telegram bot with a knowledge base from documents — an n8n template: a bot built on a knowledge base
- Express course “AI Agents: the era of electrifying routine” — $9, Creator, Make, N8N, Full plans
What to read next
The AI freelancer price list — ranges for all nine types of work and the salaries for monthly work.
The one-tool rule — why cycling through services does not turn into money.
A portfolio without clients — how to assemble the entry in one evening for any of the three tracks.
Where to find AI work — four platforms and what to do with the one carrying the most work.
AI API pricing in 2026 — what a request costs and how to work out your own costs.