Highlights
- 35% of teams have already replaced at least one SaaS tool with a custom build; 78% expect to build more in 2026 (Retool, 2026). Small business owners are now showing up in that number, not just enterprise IT.
- Sky L. of Scratch Golf created an e-commerce command center that let him cancel most subscriptions his business ran on, saving about $600 a month.
- Paul L. of TradeWeave replaced Dell Boomi, the $70,000-a-year EDI platform his trucking clients were stuck with, at roughly half the price with more features. Eight carriers run on it now.
- Every founder in this piece took a risk and built something better. Time to build ranged from a few hours to three months.
- The pattern: SaaS fits average businesses. Custom software fits yours. AI collapsed the cost gap that used to make renting the obvious choice.
Why small businesses are building software instead of renting it
The era of DIY software has arrived.
Small businesses are replacing the software they rent with software they build: 35% of teams have already swapped at least one SaaS tool for a custom build, and 78% expect to build more in 2026 (Retool, 2026). The seven businesses below are that statistic in person.
Before AI building stormed the castle of traditional software development, creating software was a gatekept skill. If you didn't know how to write code, you weren't going to be shipping software.
Now, anyone with a good idea and some time can (and is) building better, bespoke versions of the SaaS products they've been forced to rent for years. The method is called vibe coding: describe what the software should do in plain language, and AI writes the working code.
Here's a quick overview of seven business owners and founders who took the opportunity vibe coding offered and ran with it.
1. Scratch Golf: an e-commerce command center that replaced ~$600 a month in subscriptions
E-commerce · ~10 subscriptions cancelled · ~$600/mo in software gone
Sky L. and a partner bought Scratch Golf, an online store selling golf simulators. The previous owners sold for a reason that will sound familiar to anyone running a shop online: the back office setup couldn't support the volume of incoming business.
Ten-odd subscriptions kept it running. Tidio for chat. Klaviyo for email. QuickBooks for the books. Calling software and plugins on top of that. It came to about $600 a month, and the Shopify bill was double that again.
So Sky built the replacement. One tool, which he calls the command center, now pulls orders through the Shopify API, tracks vendors and product margins, handles calling and texting from the browser, logs every customer touch, keeps the full ledger, and runs the email flows. An AI answers customer questions straight out of the product catalog.
He built the storefront too, as a second project. Shopify is down to just the checkout page, and he's working on his own payment handling to close that last gap. His read on where all this goes: "Instead of having a one size fits all, you can make it fit for you. Your one size."
What he was paying for: about 10 subscriptions covering chat, email, bookkeeping, calling, and plugins, near $600 a month, with Shopify roughly double that on top
What he built instead: one command center running orders, vendors, margins, calls, texts, CRM, ledger, and email flows, plus a custom storefront
Time to build: ongoing (still shipping)
2. TradeWeave: replacing a $70,000-a-year EDI platform at half the price
Logistics · replaced a $70K/yr platform · 8 carriers running on it
Paul L. has logged over 10 years consulting on the biggest transportation management system in North America. He understands the industry, knows the software, and has wired it up for trucking clients from coast to coast.
What it costs is the story. Dell Boomi, the industry-standard EDI software that shuttles freight documents between carriers, shippers, and brokers, runs about $70,000 a year. That's before anyone pays a consultant like Paul to configure it. Then his clients started asking about modern API connections, so Paul asked the vendor. "They said, that's a $20,000 a year license."
He'd already tried Bolt.new on a smaller problem and watched it hold up. So the next question asked itself. If an AI could read freight paperwork, could it build the thing his industry had been overpaying for since the nineties?
Two days into the build, he had his answer.
TradeWeave does what the $70,000 tool does, at about half the price, with features Paul always wished the original had. They needed support ticketing, and the existing options wanted $30 or $40 a month, so they built their own. Same for licensing. Same for the tool that pushes updates out to each customer.
Eight carriers run on it now, six of them large operations coming off systems old enough to need Citrix just to open a dashboard. The entire company is Paul and his two sons.
What he was paying for: Dell Boomi at about $70,000 a year, plus $20,000 more for API access, plus per-transaction fees priced by file size
What he built instead: TradeWeave, at about half the price with more features, plus their own ticketing, licensing, and deployment tooling
Time to build: two days to a working answer, still shipping
3. Heartly and 10 more: a one-person portfolio that avoided ~€100K in development costs
Business services · portfolio of 10+ live products · ~€100K in dev cost avoided
Leo C., the founder of Heartly, runs a one-person software portfolio. Heartly itself is a digital welcome-book platform for short-term rentals, and it's one of more than ten live products they've shipped: a CRM for insurance agents, property management software, community platforms. All of it built solo, all of it running.
The method is the same every time. Find a workflow inside a small business that generic SaaS doesn't fit, describe what needs to happen the way the owner would say it, and build the bespoke version in a few hours. Then ship. The businesses on the other end get software shaped to how their industry works, not the average of the whole category.
By their estimate, hiring developers to build what they've shipped would have cost more than €100,000.
What they were paying for: N/A. The alternative they skipped was hiring a development team
What they built instead: 10+ live business products across short-term rentals, insurance, property management, and community platforms
Time to build: a few hours per product
4. NetDoctor: a telehealth marketplace built in three months
Healthcare · telehealth platform built in 3 months · 90 providers, 50 patients
Two decades in the Argentine healthcare system made Martin F. deeply familiar with how it works and what needed to improve. Argentina struggles with equity of access: most specialists are concentrated in urban centers, and many people travel for hours to meet with doctors outside of general medicine.
He knew a telehealth marketplace would close that gap. He got quotes from software companies and professional developers, and every one of them was out of reach.
Three months from his first session, he had a working platform: appointment booking, video consultations through Zoom, payments through Mercado Pago, and continuing education for the providers on it. Zoom put the whole system through a full audit against international standards before approving it for their marketplace.
He launched by posting to a single Argentine job board. Almost 2,000 views and more than 100 provider applications came back the first week. He reviewed credentials on every one, and that diligence has already caught applicants who weren't providers at all. Ninety are approved and practicing today, with 50 patients who found them on their own.
The business model is the part that makes people sit up. No subscription. Providers pay 5% on the consultations they complete, and a no-show costs them nothing.
What he was paying for: N/A. He priced a development team, found it out of reach, and shelved the idea for four years
What he built instead: NetDoctor, a telehealth marketplace with booking, video consultations, payments, and provider education
Time to build: three months to a shipped MVP
5. FlagStat: the flag football stats platform now closing a pre-seed round
Sports · flag football stats platform · closing pre-seed
Bradley M. built FlagStat because flag football had no stats tool worth renting. Leagues, coaches, and players were making do with generic sports apps built for somebody else's game: soccer templates, baseball templates, whatever a SaaS vendor happened to sell. FlagStat matches how flag football is played, and 60 days in it had customers in 8 states and 50,000 pageviews.
Flag football stats sounds like a hobby project. Bradley treats it like a company. He's closing a pre-seed round to run the same play in 10 more sports, each app built for one specific game. Owned, not rented.
What he was paying for: N/A. Flag football didn't have a stats tool worth renting
What he built instead: flag football stats platform, plus 10 more sports apps in the pipeline
Time to build: 1 to 4 weeks
6. E.L.I / EightCounty: taking lean improvement off paper
Business services · lean improvement companion · digitized paper processes
Small manufacturing and service businesses still run lean improvement on paper. Kaizen sessions, Gemba walks, five-why analyses: the insights land in notebooks and on whiteboards. The action items go into a spreadsheet and never come back.
Charles R., the founder of E.L.I (Improvement Companion), built the fix. E.L.I moves the whole flow into one connected tool. What a team learns on the floor turns into tracked work, not a lost note. The fix outgrew its first job, too. They founded EightCountyDigital Ltd around the software, and a personal tool for one small business became the company.
What they were paying for: N/A. Paper and manual tracking. No SaaS existed for this niche
What they built instead: lean-improvement companion that digitizes paper-based improvement workflows
Time to build: 1 to 4 weeks
7. Ginger Pulse: a diabetes tracker that scores meals before you eat
Healthcare · diabetes and nutrition tracking · 100+ waitlist signups in 10 days
Jai S. built the first version of Ginger Pulse in a spreadsheet, for himself.
His A1C was moving the wrong direction and his physicians kept adding medications. So he started logging what he ate against what his blood sugar did afterward, hunting for the pattern. He found it. That spreadsheet is what became the app.
The tracking software that already exists comes in two flavors. Clinical-grade tools built for endocrinologists, priced to match. Or wellness apps built around calorie counting and weight loss. Neither one is designed for the person making a hundred small decisions a day about a condition they'll have for life.
Ginger Pulse is built for that person. You photograph the meal, or say it, or type it, or scan the barcode, and it tells you what's coming before you eat instead of after. Jai's line for it: "People should know before they eat, not after they have already eaten."
The scoring is personal rather than universal. "There is no universally good food," he says. The same meal can come back green today and yellow tomorrow depending on the person, the day, and what came before it. He pulled in food databases well past the American standard ones, so Indian, Australian, and Middle Eastern cooking read right, and the app learns whether you cook with clarified butter or refined oil.
More than 100 people joined the waitlist in the first ten days. Private beta onboarding is running now.
What he was paying for: clinical tools priced for specialists, or wellness apps built for weight loss. Neither one fit
What he built instead: a diabetes and nutrition tracker that scores meals in context, before you eat
Time to build: started at the Bolt.new hackathon, still shipping toward an App Store launch
What all seven have in common
Not one of the seven is a working developer. A couple could pass for one: Jai S. has written software for twenty years, and Paul L. can code "to some extent," in his own words. They picked Bolt.new anyway, over tools they already knew. The rest described what their business needed and got working software that fits. Nobody hired a development team.
The builds range from a few hours (Heartly's founder ships whole products at that pace) to two days (TradeWeave's first working answer) to three months (NetDoctor, a full telehealth platform). Every one of them replaces something specific: an expensive SaaS bill, a manual process, an ill-fitting generic tool, a gap in the market.
The pattern isn't "small businesses are getting cheaper about software." The pattern is "small businesses are getting more specific." AI didn't lower the standard for software. It let owners raise it.
Build the software your business actually needs
Everyone in this story turned their ideas and expertise into something new.
Some had a development foundation to pull from, some didn't. Some were already business owners or entrepreneurs, some weren't. The unifying factor here is that they thought they could, and they did.
If you have a great idea you've been thinking about at stoplights, or dreamt of as you struggled with a clunky program that doesn't do half what you need it to do, let their stories give you permission to make that idea real.
