Can AI Build You a Microbusiness That Earns Money in the Background?
Alex Finn has shared an interesting demonstration of how to use AI to build a small online business that keeps working in the background.
The idea is simple: give an AI agent a useful business to run, connect it to the services it needs and let it handle much of the daily workload. That could include updating a website, researching new information, preparing newsletters and collecting sponsorship payments.
Alex calls this “agentic income”: money generated by businesses that AI agents help operate.
Having followed Alex’s work, this is the sort of practical demonstration I find particularly interesting. It takes AI beyond answering questions and writing text, and shows how it can contribute to something that actually earns money.
What Alex has built
Alex’s example is Agent Use Cases, a website that helps people discover practical things they can do with AI agents.
It addresses a familiar problem. People sign up for AI tools, experiment with them and then wonder what to use them for. A resource that brings together useful examples gives them somewhere to start.
Alex has an AI agent called Milo maintaining the business. Its responsibilities include finding new use cases, updating the website, improving its visibility in search and running an accompanying newsletter.
In the video, Alex shows $1,100 in revenue collected over the previous day from two sponsorships. One of those sponsors paid $600.
He also shows around 56,000 visitors and 123,000 page views over a couple of days, alongside approximately 2,000 subscribers to the website’s newsletter.
The business model is straightforward: create a useful resource, attract people interested in the subject and sell sponsorship space to companies that want to reach them.
Start with a problem people need solving
Alex’s starting point is choosing a small, specific problem.
For the demonstration, he asks his agent to suggest business ideas based on subjects his audience already follows: AI, technology, gaming and building software.
The suggestions include a service that explains coding errors, help with writing launch posts, a directory of AI tools and ready-made agent templates.
He chooses an AI tool directory aimed at helping people answer a practical question: “Which AI tool should I use for this task?”
That is an important part of the approach. You do not need to invent the next enormous technology company. You need something useful that serves a clear purpose and is manageable enough for an agent to maintain.
A focused directory, a specialist information service or a resource for a particular community could all fit that general model.
Give your AI agent a dedicated role
Alex creates a separate agent for the demonstration and gives it responsibility for the new business.
He describes the agent as its “CEO”. In practical terms, that means giving it a defined job: help develop the idea, build the website and organise the ongoing work.
The initial conversation establishes what the business will do, who it will serve and how it could make money.
The agent then builds a first version of the directory, called Build Stack, containing 33 tools organised into categories.
Alex makes the point that the first version does not have to be perfect. The priority is getting something useful built and available, then improving it.
That is one of the most appealing parts of the demonstration. An idea can become a working resource quickly, allowing you to spend more time assessing its usefulness and less time wrestling with the initial build.
Tell the agent how to keep the business running
Building the website is only the beginning. Alex then sets out what the agent should do regularly.
Those responsibilities include:
- Finding new tools and releases.
- Updating the website’s information.
- Checking for bugs.
- Monitoring feedback.
- Improving search visibility.
- Starting and maintaining a newsletter.
- Developing suitable ways to monetise the business.
The agent creates scheduled routines for daily maintenance and weekly progress reports.
This is what makes the approach different from asking AI to build a website and leaving it there. The agent has an ongoing workload, with instructions to keep the resource useful and report back when something needs attention.
Alex’s existing agent also sends him updates when sponsorship payments arrive or improvements are being made.
The aim is to reduce the amount of daily administration the owner needs to handle, while keeping them informed about the business.
Connect the tools it needs
For an agent to operate a business, it needs access to services that allow it to take action.
Alex uses:
- Vercel for website hosting.
- GitHub to store and manage the code.
- Resend to distribute the newsletter.
- Stripe to collect payments.
These connections allow the agent to move beyond making suggestions. It can work on the website, send emails through the appropriate service and provide a payment route for sponsors.
Alex also mentions connecting Google Search Console to help monitor the website’s search performance.
The exact setup will depend on which agent you use and the integrations it supports. You will also need to account for the costs of AI usage, hosting, a domain, email delivery and payment processing.
One practical correction to the video’s hosting advice: although Vercel offers a free Hobby plan, its published rules restrict that plan to personal, non-commercial use. A business earning revenue needs a hosting arrangement suitable for commercial activity.
How the money comes in
In Alex’s example, the income comes from sponsorships on the website and in its newsletter.
For an AI directory, the potential customers are companies that want their tools to reach people looking for useful software.
A sponsor can pay for a featured placement, with Stripe providing the payment process. The newsletter gives the sponsor another opportunity to reach an interested audience.
That makes the email list a valuable part of the business. It helps bring readers back, gives the agent a regular publishing task and creates additional space for paid promotion.
Paid placements should be clearly identified so readers understand which listings are sponsored. A useful directory depends on people trusting its recommendations.
Getting people to visit
Alex is clear about the importance of distribution: having a way to get your project in front of people.
He has built a substantial audience on X and uses it to promote Agent Use Cases. That gives him a strong starting point when launching something new.
His advice is to begin building an audience of your own, even if it is small. Share useful information, participate in relevant conversations and introduce your business when it offers something people will find helpful.
For anyone who already runs a website, newsletter, channel or community, this is particularly interesting. You may already understand your audience’s recurring questions and have somewhere to introduce a resource that answers them.
Search visibility matters too. Alex instructs his agent to improve the website for search engines and make its information easier for AI chatbots to understand and reference.
Those efforts can help people discover the resource, alongside the owner’s own promotion.
How hands-off can it become?
The attraction is having a business that continues doing useful work without requiring your attention throughout the day.
Alex’s demonstration shows how an agent can take on a considerable amount of that workload: researching, updating, maintaining, preparing newsletters and handling parts of the sponsorship process.
The owner still chooses the direction, connects accounts and makes decisions when needed. In the demonstration, the agent asks for details such as a real contact email and says it will seek approval before certain communications or payment-related actions.
Once those foundations are in place, however, much of the recurring work can run in the background.
That is the opportunity I find interesting here. A small online business has traditionally meant taking on another list of jobs. With an AI agent handling many of those jobs, it becomes more practical to develop an additional resource alongside your existing work.
Why this is worth a look
Alex’s example brings together a useful idea, a straightforward income model and an AI agent capable of handling much of the operation.
The technology is impressive, but the underlying business principles remain familiar: understand your audience, solve a problem and offer something worth returning to.
If you already know a subject well, the starting question is simple: what small resource would people in your audience find useful enough to keep using?
Alex’s video shows how AI could help you build that resource and keep it working while you get on with the rest of your day.