Online creators claim that AI-generated influencer accounts can produce large revenue with little effort. This article documents a seven-day test of that claim. The test built three AI influencer accounts, selected one, and sold a digital product. It reports the process, the results, and the full cost and profit calculation.

Claims Under Review

Promotional content about AI influencers cited the following figures. None were verified in this test.

• Creators were making thousands of dollars using Claude to build AI influencers, without showing their faces.
• One creator claimed $100,000 USD in revenue over 30 days from AI creators.
• Other claimed results included over $2,000 USD in sales,$200,000 USD in profit, and over $103,000 USD in earnings.

Challenge Parameters

The author set the following rules:

1. Total startup cost must stay under $99 USD.
2. The author could not use their own face or voice.
3. The first AI influencer had to be created on the first day.
4. The goal was to earn at least $1,000 USD within seven days.

The author stated that failure to reach the goal would result in a $1,000 USD payment from the author to a reader.

Market Research

The author reviewed existing AI influencer accounts. The findings were as follows:

• Lil Miquela: Over 2 million followers. The account was reported to earn over $11 million USD per year through brand deals with Prada and Calvin Klein. The character appeared at the Met Gala and the Grammys and is positioned as a supermodel.
• Mother Satori: Presented as a US-based holistic expert and as a 108-year-old. The account promotes a gut cleanse detox drink at the end of each video and receives thousands of views daily.
• Relationship account: A character posting lifestyle photos and podcast-style reels about dating.
• Woodworking account: A character presented as a patriotic British craftsman producing wooden goods. The author stated that the account misled customers. The author also stated that the account was likely generating significant income.
• AI granny account: Over 2 million followers.

The author identified health, wealth, and relationships as the three niches that appeared most profitable. Two of these niches already had proven accounts. None of the reviewed AI accounts covered money or wealth. The author identified this as a gap in the market.

Niche Selection and Character Creation

The author created three AI influencers, one for each niche. The author used Claude's free plan to generate character profiles. Each prompt requested a name, a backstory, a distinct look, and a unique selling point.

Image Generation

Tool: Higgsfield. The author selected the Ultra plan because it supports over 100 videos. The plan's listed price is $99 USD. The author counted the full price in the cost analysis.

Setup steps:

1. Create an account using email.
2. Connect Higgsfield to Claude. In Claude, the author opened Customize, selected Connectors, chose Add, and then added a custom connector using the Higgsfield link.
3. Ask Claude to write the image prompts.

Image specifications:

• Full-body and close-up angles.
• White background.
• Realistic skin texture.

Results:

• Amos: The first images appeared younger than the age in his profile. The author requested older-looking images, and the revised version was accepted. Scenes included a porch, a farmhouse kitchen, a vegetable garden, and a yard.
• Grandma Vivian: Images show sunglasses, jewelry, and a fur coat. Locations include a private jet, a sports car, a drawing room, and a terrace.
• Sienna: Images show a stylish, realistic appearance. Locations were generated to match her profile.

Content Scripts

The author used Claude skills to generate content. A Claude skill is a reusable tool with built-in instructions. The author downloaded free skills online and uploaded them through Customize, then Add, then Upload skill.

• Content engine skill: Generated a list of content ideas for each character. The author selected three scripts for Sienna and produced scripts for the other two influencers.

Voice Generation

1. In the Higgsfield audio tab, the author uploaded a character image for Vivian.
2. The author entered a sample script: "Hi, my name is Vivian."
3. Claude generated a detailed voice prompt. The author pasted it into Higgsfield and generated the voice.
4. The same process was repeated for Amos and Sienna.

The author described Vivian's voice as similar to a former schoolteacher's voice.

Video Production

Tool: Higgsfield Cinema Studio.

Settings:

• Aspect ratio: 9:16, for vertical short-form video.
• Quality: 1080p.

Process:

1. Create a new project and name it.
2. Start with silent B-roll clips. These allow text overlays and test output quality before the character speaks on camera.
3. Use a Claude skill to convert a simple idea into a detailed video prompt.
4. Reference the Sienna character sheet image to keep the character consistent across videos.

The author reported that some generated clips were convincing. Sample outputs included an early-to-bed proverb with realistic coffee steam, a claim that beans are a longevity food, relationship advice about a friend who disappears when she gets a boyfriend, and property investment advice presented by an older woman.

Launch

The author launched accounts on Instagram only, due to the time limit. TikTok and YouTube Shorts were identified as possible later channels.

Amos account:

• Category: Health and wellness.
• Username: dot old ways.
• Profile picture: Amos with a lamb.
• Bio: "101 years old, simple habits for peak energy, focus and longevity." The bio also states "AI grandpa" to disclose the account's nature.

The author created accounts for all three influencers and planned to post 10 videos on each.

Results

The author compared performance after several days:

1. Amos: First place. The author named Amos the winner and moved all further effort to his account.
2. Sienna: Second place. Her numbers were described as strong for a new account.
3. Grandma Vivian: Third place. The author had expected her to perform best.

Sienna's comments included criticism of her sheen dress and her $3 USD Frappuccino. Another commenter defended her choices. The author also noted that viewers asked where her clothes came from.

After the author stopped focusing on Amos, Vivian's account grew. One video received over 10,000 views, another received 20,000 views, and a third received over 250,000 views. The author stated that the choice of Amos was a mistake. The author also stated that none of the comments on any account complained about AI use.

Monetization Options

The author evaluated six options:

Digital Product

The author selected a digital product because the author judged that generic ebooks are available free through search and AI tools. A product that is structured, interactive, and guides users through a process was considered more likely to sell.

Product creation:

1. The author used the Claude digital product skill. It generated three product ideas with reasons customers would buy each one.
2. The author selected a 30-day challenge based on Amos's approach, with one habit per day.
3. Claude built the workbook. Each day has a named rule and a tick box. The workbook includes a tracker and bonus content, including Amos's mother's dinner recipes.
4. The author added an image of Amos to the cover.
5. The author reviewed the content for logical soundness, safety, and signs of AI-generated text.

Sales setup on Stan Store:

• Free 14-day trial.
• Product type: digital product.
• Style: Callout.
• Thumbnail: a sheep image that matches the Instagram profile.
• Title: "30-Day Energy Reset."
• Button text: "Start the Old Ways Challenge."
• List price: $15.99 USD.
• Discounted price: $12 USD.
• Payments: connected to PayPal.
• Distribution: The product link was added to the Instagram bio. The author tested the link before launch.

Financial Results

The author reported 23 sales of the workbook at $12 USD each.

The author counted the full $99 USD Higgsfield price because the discount was limited to the first 5,000 users. The PayPal fees were not part of the startup budget. They were deducted from sales.

Assessment

The author concluded that the following points applied:

• The $1,000 USD target was not reached. The reported profit was$157.48 USD.
• The seven-day period was short. The author stated that Vivian's account gained traction after the winner was chosen.
• The method could be extended to other platforms and monetization options. The author did not test these.
• The author stated reservations about AI-generated content on social media, citing the value of connecting with real people.

The author stated that the test showed AI influencers can be created and monetized with minimal setup and a budget of $99 USD.