The Free Software Behind a New Kind of Side Hustle

Picture the owner of a four-chair dental practice in a mid-sized American city. Every month she pays an agency to run her Google ads and "do SEO." Every month she gets a PDF full of charts she doesn't read. She has no idea whether new patients are finding her, or whether the practice across the highway is quietly taking them.

Now picture someone on the other side of that problem. Maybe a recent graduate, a laid-off marketer, or a retail salesperson with evenings free. He has never written software. But he has a laptop, an AI assistant that writes code on request, and a free program downloaded from the internet that does most of what her agency's expensive tools do.

The distance between those two people, one with a problem and money, the other with cheap tools and time, is the premise of one of the fastest-growing genres in online business advice. It holds that the most valuable software of the moment is increasingly free, and that the real opportunity lies in the unglamorous work of turning it into something a busy owner will pay for.

It is a seductive idea. It is also, in important ways, correct. But the version sold in videos and course funnels tends to skip the parts that decide who actually makes money: the licenses, the competition, the sales grind, and the question of who maintains the free software everyone is building on.

A short history of free

Free software did not begin as a business strategy. In the 1980s and '90s it was closer to a movement: programmers who believed code should be shared, studied and improved in public. The operating system Linux, begun by a Finnish student in 1991, became its best-known product.

What happened next surprised almost everyone, including many of the movement's founders. Free software turned into the invisible foundation of the commercial internet. Linux came to run most of the world's servers. WordPress, a free publishing system, came to power a large share of all websites. Companies that once fought open source ended up among its biggest contributors, because sharing the plumbing was cheaper than everyone building their own.

For decades, though, open source mostly helped people who could program. Using a project from GitHub, the code-sharing site now owned by Microsoft, meant reading setup instructions full of terminal commands, fixing whatever broke, and knowing enough to judge whether the code could be trusted. For a non-technical person, "free" software often cost more in frustration than a paid app cost in dollars.

Two things have changed since then. The first is supply. The boom in artificial intelligence has set off an explosion of open-source projects that copy, and sometimes beat, the features of expensive commercial tools. The second, less noticed, is access. AI coding assistants such as Anthropic's Claude Code can now read a project's instructions, install it, and fix errors in response to plain-English requests. The skill needed to run free software has fallen almost as fast as its price.

Put those together and you get a new kind of arbitrage. The software costs nothing, the know-how is rentable for a few dollars, and the buyers are the millions of small businesses who will never do any of it themselves.

The playbook

One widely watched version of this pitch arrived in September from Corey Ganim, an entrepreneur who runs a paid online community for people building AI service businesses. His video, "5 Github repos that print $$$," drew more than 200,000 views in under two weeks.

The format will be familiar to anyone who has spent time in the online-business corner of YouTube. There are five tools, each paired with a target customer, a price and a monthly revenue projection. The advice is brisk and specific. Let the AI assistant handle the technical setup. Choose a narrow type of customer. Charge a setup fee plus a monthly retainer. Then find buyers with a people-search tool that pulls verified contact details from dozens of data providers.

The core of the pitch is a line that could serve as the motto of the whole genre: "We're not selling AI. We're selling a specific outcome."

That distinction matters more than it first appears. Small-business owners have been pitched "AI" relentlessly for three years, and many have grown wary of it. What they have not grown tired of is getting more customers, answering fewer repeat questions, or closing deals faster. The playbook's cleverest move is to hide the technology entirely and sell the result.

The second clever move is in the sales method. Instead of cold pitches, it recommends leading with proof. Email a law firm the three worst findings from a free audit of its website. Send a podcaster three finished clips from last week's episode. Ask a sales director for their last proposal and rebuild it while they watch. In each case, the free tool does the work that earns permission to ask for money.

It is a good sales strategy. It is also one that a very large number of people just learned at the same time.

Five tools, up close

The projects in the video are real, active and, by the standards of GitHub, popular. Stars, GitHub's version of a bookmark, are an imperfect measure, but they show where developers' attention is going.

Scrapling: watching the competition

Scrapling, with about 69,000 stars, is a web scraper, software that reads websites and pulls out data. Its distinguishing feature is resilience. Most scrapers break the moment a site redesigns a page. Scrapling tries to find the same price or stock label again, and it includes modes designed to get past common bot-blocking systems.

The proposed service is competitive intelligence for small online stores: track five rivals' prices and stock levels, log them in a spreadsheet, and alert the owner the moment a competitor cuts a price or sells out. Large retailers pay enterprise vendors for exactly this. A Shopify brand doing a few million dollars a year usually has nothing of the kind.

Dify: the handbook that answers back

Dify is among the most-starred AI projects on GitHub, with more than 150,000 stars. It lets users build chatbots and automated workflows on a visual canvas and connect them to nearly any AI model. It can be self-hosted or run on the company's paid cloud, whose Professional plan costs $590 a year.

The proposed service is a private chatbot trained on a company's procedures and handbook, sold to firms with high turnover such as home-services companies, property managers and staffing agencies. The pitch rests on a real, everyday cost: the hours owners and managers spend answering questions whose answers are already written down.

OpenSEO: the scorecard

OpenSEO, with about 21,000 stars, describes itself as "an SEO tool for the people." It covers keyword research, rank tracking, backlink analysis, site audits and a newer category, visibility in AI-generated answers. Rather than charging a flat subscription, it passes through the cost of search data bought per request from a provider called DataForSEO. Its hosted version costs $10 a month plus a markup; self-hosting skips even that.

The proposed service is a $499 audit followed by a $199 monthly scorecard for local businesses such as dentists, law firms and HVAC companies. The video calls it the most beginner-friendly option of the five, and the economics suggest why.

OpenShorts: the clip machine

OpenShorts, with about 5,000 stars, takes long videos and finds the moments most likely to perform as short clips, using Google's Gemini AI to read the transcript. It crops the footage to vertical format, follows faces, adds word-by-word captions, and can post directly to TikTok, Instagram and YouTube.

The proposed service is a monthly package of 30 to 40 finished shorts for professionals who already record long content, such as podcast hosts, realtors, coaches and advisers, at around $400 a month. The software is a commodity; the product is taste. Someone still has to decide which clip is actually good.

Presenton: the proposal factory

Presenton, with about 10,700 stars, generates presentations that can be edited in PowerPoint. It builds templates from a company's existing decks and offers an API, a way for other software to request a deck automatically. It works with most major AI models and can run on a company's own machines.

The proposed service is same-day proposals for sales teams, built from a short form or even a recording of the sales call. It carries the highest price in the video, a $2,500 setup fee plus $200 a month, justified by the argument that faster proposals close more deals.

The economics, examined

On paper, the margins are extraordinary. Here are the video's own figures, each assuming a small book of clients:

Against that, the stated running costs are close to trivial: $10 a month for SEO hosting, $40 for scraping proxies, a few cents of AI usage per video. By any traditional measure of a services business, the cost of goods sold is effectively zero.

But cost of goods is rarely what sinks a small service business. Three other numbers matter more, and they get less airtime.

The cost of finding clients. The playbook suggests sending hundreds, even a thousand or more, cold emails to book a handful of calls. Each contact costs only a few cents to look up, by the video's estimate. The real cost is time: researching prospects, writing personalized proof, following up, sitting on calls that go nowhere. Getting ten clients who each pay $199 a month could easily take months of steady outreach.

The cost of keeping them. Monthly retainers depend on renewals, and small businesses cancel services readily when money is tight or results feel abstract. A client who pays $199 a month for four months and then leaves is worth less than $800. The math only compounds if the service keeps proving its value, month after month.

The cost of your own time. The video's estimate of "about 10 hours a month" to serve ten clients covers the work of running the tools. It does not include sales, billing, client emails, fixing broken setups, or the unpaid learning curve before the first sale. Self-employment also carries its own taxes and no benefits.

None of this makes the model a bad one. A few thousand dollars a month in recurring revenue is a meaningful side income, and service businesses with near-zero software costs are genuinely unusual. But the fair comparison is not "free tool versus $200 retainer." It is "everything it takes to win and keep a client" versus what that client pays over the life of the relationship.

The customer's side of the table

It is worth returning to the dentist, because her perspective is the one most often missing from videos about how to sell to her.

She does not care whether her reporting comes from Semrush or from a free project on GitHub. She cares about three things: whether the information is accurate, whether it tells her what to do, and whether the person sending it will still be around next year. Price matters, but mostly as a signal. A $199 report she trusts is worth more to her than a $49 report she doesn't.

That has uncomfortable implications for the "free tool" pitch. If the provider leads with how cheap the software is, the owner may reasonably ask why she is paying for it at all. The playbook's instinct to hide the technology is correct. What replaces it has to be expertise, reliability and, above all, clear explanations.

The same logic runs through every service in the video. A store owner does not want a spreadsheet of competitor prices; she wants to know which three products to reprice this week. A property manager does not want a chatbot; he wants new hires to stop interrupting him. A sales director does not want an AI deck generator; she wants the proposal sent before the prospect cools off.

The providers who understand this will write fewer reports and give more recommendations. They will also be the hardest to replace, because what they sell is judgment. Judgment, unlike open-source code, can't be downloaded.

The fine print

"Free" in open source means free to use, study and modify, under conditions. Those conditions, set out in each project's license, rarely appear in a video thumbnail. They matter a great deal to anyone building a business.

Licenses are not all alike. Scrapling uses the permissive BSD license. OpenSEO and the core of OpenShorts use the MIT license. Presenton uses Apache 2.0. All three license types allow commercial use with few strings attached. Dify is different. Its license is based on Apache 2.0 but adds conditions: without written permission, the code may not be used to run a multi-tenant service, and the logo and copyright notices in its interface may not be removed. A consultant who puts ten clients on one installation of Dify could be out of compliance without realizing it. OpenShorts, too, keeps one folder of its code under a separate commercial license.

Licenses can change. In recent years several well-known projects, including HashiCorp's Terraform and the database Redis, have moved to more restrictive licenses as their companies sought revenue. A business built on today's terms is betting, at least a little, on tomorrow's.

Scraping sits in a legal gray zone. American courts have generally been reluctant to treat the collection of publicly visible web data as computer hacking. The long-running dispute between the data firm hiQ Labs and LinkedIn is the best-known example. But the same case showed that a website's terms of service can still create liability, and hiQ ultimately lost on those contract grounds. Anyone selling competitor monitoring should stick to public pages, respect rate limits, and read the terms of the sites they watch.

Privacy travels with the data. A handbook chatbot means a company's internal documents flow through whichever AI provider powers it. A shorts service means handling someone else's recordings and posting under their name. Both call for written agreements about who owns what, where the data goes, and what happens when the contract ends.

Free tools often have paid parts. OpenSEO's data comes from a paid provider. OpenShorts relies on outside services for its AI, voices and posting. Every added client raises those bills, and any of those providers can change its prices or terms.

Who really gets paid

There is an old observation about gold rushes: the most dependable fortunes went to the people selling picks and shovels. The online-business version adds a third trade, selling maps.

Mr. Ganim's video ends, as many in the genre do, with an invitation to a paid program. Its description promises members they will close a first AI-services client within 90 days or get their money back. That does not make the advice in the video wrong. Much of it is sensible, and the tools are real. But it is worth knowing that in this economy the most reliable revenue often comes from teaching the business, not doing it.

The other overlooked party is at the far end of the chain: the people who write the software. Most open-source projects are maintained by small teams or single volunteers, often unpaid or thinly funded. The risks of that arrangement became vivid in 2024. An attacker spent roughly two years winning the trust of the overworked volunteer maintainer of xz Utils, a small compression library found in countless Linux systems, and slipped in a hidden backdoor. It was caught almost by accident.

A service business that runs on someone else's free code has a direct interest in that code staying healthy. The most responsible operators report the bugs they find, improve the documentation, and, once revenue allows, sponsor the projects that pay their bills. It is both the decent thing to do and a sensible hedge.

What lasts

Small businesses have always needed help being found online, onboarding people, and winning customers. What has changed is not the need but the price of the tools that meet it, which has collapsed toward zero, and the skill needed to use them, which AI has sharply reduced.

That shift is real, and it opens a door for people who could never have started a software-based service a few years ago. A salesperson with good instincts, a marketer with a feel for local businesses, or a video editor with taste now has access to the same tools as a well-funded agency.

But open doors attract crowds. When the software is free and the instructions are public, the tools stop being an advantage. What's left to compete on is what has always separated durable service firms from short-lived ones: expertise in a specific kind of customer, reliability, honest reporting, and the ability to explain results in plain English.

The people who succeed with these tools will probably not think of themselves as running an AI hustle at all. They will think of themselves as the local-search firm that dentists trust, or the team that gets contractors' proposals out the same day. The free software will simply be how they do it, the way a carpenter doesn't advertise the brand of their saw.

Sources

• 5 Github repos that print $$$, Corey Ganim, September 16, 2026
• Project pages for Scrapling, Dify, OpenSEO, OpenShorts and Presenton, checked September 28, 2026
• Dify license and pricing
• Background on Linux, the hiQ v. LinkedIn case, license changes and the xz Utils incident is drawn from widely reported public events