How an AI blogger earns: four income models

How an AI blogger earns: four income models

AI characters are the second most in-demand subject, and almost everything written about it stops at the technology. Below is the money: four income models with honest timescales and ceilings, including the one that does not work the way it is promised.

AI characters are the second most in-demand subject, and almost everything written about it stops at the technology. Below is the money: four income models with honest timescales and ceilings, including the one that does not work the way it is promised.

In short, if you are pressed for time

An AI blogger has four income models, and they differ sharply in timescale: advertising on your own character, a character for somebody else's business, UGC for retail, and selling a trained asset.

The fastest to pay is somebody else's audience, not your own: a character for another business, where the client pays for a result rather than for followers.

The biggest demand is UGC: the "AI creators" category is in 86% of the orders in the bot feed.

Selling an asset is narrower than advertised: the buyer must be able to generate on their own, otherwise they need ready-made materials, which is a different model.

The rule of choice is simple: need money this month, UGC; building an asset for a year, your own character.

Cost is visible before you press the button: an image $0.12–0.25, animating a frame from $0.17, Lip Sync $0.12–1.50, a whole video from generations about $2.5–3, and the montage estimate is shown before launch.

"I started an AI blogger" — and then silence

AI characters are the second most requested subject in my survey: 45.2% are interested in exactly this.

And almost everything written about it stops at the technology: how to generate a face, how to hold it across a series, how to make it speak.

About money, almost nothing. A person masters generation, starts an account, publishes handsome frames and gives up two months later, because it is unclear what it is for.

Meanwhile there are four income models here, and they differ fundamentally in timescale, effort and ceiling. Let us go through each honestly, including the one that barely works.

Model 1. Advertising on your own character

The most obvious and the slowest.

How it works. You grow an account with a virtual character, build an audience and sell placements.

What is needed: a consistent character who does not change from frame to frame, regular content and time. A lot of time.

The honest timescale. An audience accumulates over months. First money comes no sooner than the account becomes noticeable, and that is not a question of picture quality.

The main constraint nobody mentions: an advertiser needs figures rather than beauty. A virtual character with a thousand followers is worth the same as a live blogger with a thousand followers — that is, almost nothing.

Who it suits: those prepared to run a project for a year and not expecting a return in the first months.

Who it does not suit: most people. This is not a way to start earning, it is a way to build an asset.

Model 2. A character for somebody else's business

The most underrated and the fastest to money.

How it works. You do not grow your own account — you make a consistent character for somebody else's brand. A face the company uses in advertising, listings and social media.

Why people buy it. A business has three alternatives: shooting with a model — expensive and from scratch every time; using stock — impersonal and identical to competitors; working with no face — recognisability is lost.

A trained character solves all of it at once: the same face across any number of materials, with no shoots and no rights to somebody else's person.

What is delivered to the client: the trained character, a set of starter materials, instructions for use and the rule that further material is generated in the same form.

The order of prices. In our bot's feed of 5 986 jobs over four months: an image per piece $12–35, and a set of marketplace listings $245–730. A character for a brand is not per-piece work; it is closer to the second category.

Who it suits: those who want money this month rather than in a year.

Model 3. UGC content for retail

The largest area of demand of the four.

How it works. A brand needs a clip "as filmed by an ordinary customer": the character holds the product and speaks to camera. Hook, demonstration, call to action.

Why this specifically. Studio advertising has stopped working in feeds — people recognise it and scroll past. A live review works. And filming live reviews with real people is expensive, slow and does not scale.

How large this is: in our bot's feed the "AI creators" category appears in 86% of jobs — the largest. UGC is broken out as a separate category in another 4%.

What is needed: a character with rights, the product in a photograph, and the ability to hold a series. Plus an understanding that the clip should look unprofessional — that is the whole point of the format.

The order of prices: a short clip $18–45, turnkey AI video $245–550. The difference is that in the second case what is sold is not a clip but a set for a campaign.

Who it suits: those who can already make a talking character and are ready to work in series.

Model 4. Selling an asset

The narrowest, but with interesting mechanics.

How it works. You train characters and styles and sell them as a finished asset: not work but a tool.

What that can be: a trained character for a specific niche, a visual style (LoRA), a voice clone.

An honest limitation: the market is narrow. The buyer has to be able to generate himself — otherwise what he needs is not an asset but finished material, that is, model 2 or 3.

Who it suits: those working with other contractors rather than with end clients.

Which income model to choose

A simple rule by timescale.

Need money this month — model 3, UGC. The largest demand, the lowest entry threshold, a clear fee.

Need a steady flow — model 2. One client with a trained character comes back for new material constantly. In the scrape 23% of clients are explicitly looking for somebody long-term.

Building an asset over years — model 1, your own blog. With your eyes open to how slow it is.

Working inside a professional community — model 4.

And most of all: do not try everything at once. The one-tool rule applies here as everywhere: one income model until the first payment, the rest later.

What an AI blogger needs technically

Three things, without which none of the four models works.

A consistent character. Not a reference, but a trained hero (a LoRA) who is placed into any generation with an unchanged appearance. You train it in the Gen AI section, in your own account. Without it a series cannot be assembled, and all commercial work is series.

Rights to the face. Someone else's face cannot be used in any form. There are three legitimate sources: your own photo, your own trained character, or a character generated by a model — not stock people and not other people's faces.

A voice, if the character speaks. A voice clone, also trained in Gen AI, or a generated voice-over plus lip sync.

Photos for training the character

The photo set decides whether the hero is as recognisable in the hundredth video as in the first. A few rules that save paid attempts.

Different angles and expressions: front, three-quarter, profile, smiling and neutral.

Different light and backgrounds, but without heavy shadows on the face. The same wall in every frame will teach the model to confuse the hero with the interior.

No filters, masks or retouching, and no glasses or headwear in most frames. If glasses are in one photo and not in the rest, the result will drift.

No AI images in the training set. A generated face with flaws becomes fixed in the character for good.

Good sharpness, no group shots and no duplicate frames. Fewer but different is better.

Cost and fee

Generation is paid from your balance as it happens, the price is visible before you press, and money for what was not done is returned to the balance. Money on the balance does not expire.

The order of magnitude: an image $0.12–0.25, animating a frame from $0.17, Lip Sync $0.12–1.50, music and voice-over $0.12–1.80.

With a fee of $18–45 for a short video and $245–730 for a pack, the cost is a few percent. Put it in the estimate the way a print shop puts in paper — a contractor who pays from their own pocket and does not count works at a loss without noticing.

How it is produced on the platform

Two roads, and they do not replace one another: the video is assembled in one place, regularity is provided in another.

Road 1. The videos themselves: Cerebrum and the AI SKILLS connector for Claude

Your character and voice clone are trained in Gen AI. After that, videos with them are assembled as one task in Cerebrum — a chat where you describe the task in plain words, the right models are chosen automatically, an estimate is shown before any paid step, and the run starts only after your agreement. Payment from the balance as it happens, files are attached with the paperclip or by link.

You can do the same from an ordinary Claude chat through the AI SKILLS connector for Claude (Full plan): Settings → Connectors, address https://aiskills.team/mcp, sign in with your aiskills.team account. Input files are passed as links: Google Drive with access for "anyone with the link", a link to a previous generation from "My generations", or a direct link.

The montage pipeline places your character and voice into the video and adds subtitles, graphics and sounds. A 9:16 cover and a publishing pack (captions, hashtags) are included free.

For the four income models there are ready-made pipeline skills; they are free and run in Cerebrum and through the connector:

"Storytelling with a consistent character" — a series of stories with one hero: the basis of models 1 and 3.

"UGC ad from a product photo" — a "filmed by a buyer" video from a product photo: model 3.

"From selfie to ad" — an advertising video from your own photo: good for a portfolio and for model 2.

"Video from idea to result" — the whole path from concept to finished video.

The skills catalogue, category "Content generation": aiskills.team/claude-skills.

Advertising and cinematic videos are made from starting frames: photos are generated first, a storyboard is built from them, each frame is animated with the video model you choose, and the scenes are joined with music, voice-over and subtitles. You can attach photos of a client's products and get an advertising video for a shop.

Cerebrum has no schedule and no autorun: every video is a separate task. For regularity you need the second road.

Road 2. Regularity: n8n and Make templates

Publishing on schedule, collecting ideas, replying in direct messages and analytics are not the video but the process around it. Automation templates cover it: more than 3,000 ready-made n8n workflows and Make scenarios. Descriptions are open to everyone, files of ordinary templates are available on the Creator, Make, N8N and Full plans, advanced ones only on Full (or for a one-off $9, or for 1,000 experience points).

A template is a JSON file: download it, import it into n8n or Make, connect your own accesses (social networks, spreadsheets, keys), switch it on. Schedules, webhooks and branching are native there and run on their own.

An honest note on the limits: an n8n or Make template does not call Cerebrum. It generates texts and images itself through the services connected in it and publishes. So the typical combination is: you assemble the quality video in Cerebrum or through the connector, while publishing, ideas and records are handled by a template.

What is worth a look for an AI blogger:

Publishing on schedule

Instagram auto-posting with AI-generated content (n8n): on a schedule it takes an idea from Google Sheets and publishes a post with a caption.

Cross-posting videos from Google Drive to TikTok, Instagram and YouTube (n8n): you put a finished video in a folder and it goes to three platforms with descriptions.

Publishing videos to TikTok and Instagram with error alerts (n8n): the same for two platforms, and a failure sends a message to Telegram.

Ideas and content plan

YouTube ideas from competitor analysis (Make): collects data about competitors and puts video ideas into a spreadsheet.

Talking to the audience

Instagram auto-replies through an AI agent (n8n): receives incoming messages and replies on behalf of the agent.

Analytics

Daily Instagram statistics to Google Sheets (Make): profile statistics into a spreadsheet every day, so you can see whether the account is growing.

For the "virtual ambassador" model

Instagram account automation for a virtual influencer (n8n, advanced: Full, $9 or 1,000 XP): runs a digital character's account from the image to video posts, with approval at every step.

The courses "Automation in Make" and "Automation in n8n" are included in the Make and N8N plans (each its own course) and in Full.

What this changes in the four income models

Your own blog. You assemble videos in batches and hand publishing to a template on a schedule. Time goes on ideas and quality, not on manual posting.

UGC videos for brands. A series of videos from the client's product photos is assembled with the "UGC ad from a product photo" skill. Instead of piece work you can offer a monthly pack of videos, and keep publishing and the statistics report running in a template.

Virtual ambassador. The client's permanent hero is trained once, and after that materials come out with the same face and voice. Account management is automated with templates.

Selling characters. A trained character with a voice is a digital asset assembled once. Before selling, check that you have the right to hand over the materials it was trained on.

Instead of $500 for running one account you can offer a pack of short videos, because production is partly automated.

The product part. Freelance 2.0, prompt No. 38 "Monetisation system", Basic plan: a four-step product ladder and a three-month launch plan.

And an honest caveat

There is no fully unattended automation, and that is deliberate. Check a video before it is published, read auto-replies and captions before switching them on, and in the Freelance 2.0 system the generators do not write text until you have chosen a hook, and do not assemble a carousel without confirming the topic. The approval points are built in on purpose.

Registration on the platform is free and opens three days of the Basic plan, no card needed.

Where to start today

One action. Open the order feed and find three listings with the word UGC or "product review".

Look not at the price but at what they ask for: length, format, how many videos. It usually turns out they need fifteen seconds, vertical, with a person and a product — and that is exactly what can be done today.

Choose one income model out of four. Not two.

What to take on the platform for this task

What to read next

AI account: what gets banned — the limits for those running their own.

Face rights in advertising — the condition for commercial work.

A video from a product photo — the biggest of the four models.

An AI freelancer's price list — what such work pays.

AI video generation in 2026 — the guide to video models in full.

Also on AI SKILLS