Starting a business used to take me 12 months. Now it takes a weekend.

The shape every business that worked had in common

Every business I've started has made money. Not because I'm gifted at business. Because when I look back at the five I've built, the ones that worked all share the same shape: a real problem someone was already paying to solve, a small first version of an answer, and a payment link that worked before I had anything polished to show.

Get Psyched, my first one, started in COVID lockdown with a Facebook page, a handful of testimonials, and a Stripe link. No website, no brand, no logo. Students messaged me on Facebook, we sent files back and forth in DMs, and I tutored psychology Honours students from my bedroom. The problem was real enough that nobody asked about the website. Five years later that business still pays me, including the stretches where I barely touched it.

TutorPro, the most recent one, started the same way. One early customer paid for access before the app had a proper login screen. I wanted to see if I could turn the AI tools I'd built for Get Psyched into a subscription product of their own.

By then I had three separate businesses living inside one parent company — the tutoring service, the online courses, and TutorPro — each with its own product offering, all sharing the same audience and brand. Building all three solidified for me what actually goes into making a business support itself without you there to man it.

Why I always build on the side of a 9-to-5

People sometimes treat 'on the side' like a smaller version of starting a business. I think it's actually the harder version, and it's the one that's taught me the most.

When you build something on the side, you can't lie to yourself about whether the problem is real. If the work feels like dragging a boulder up a hill after a full day at your day job, the market is telling you something. The exhaustion is a feedback signal. The ideas that flow on a Sunday morning before you've had coffee, where you keep getting paid even when you don't push, those are the ones that survive.

The other thing 'on the side' does is keep your runway honest. You're not gambling rent on a hunch. You're testing a hunch with weekends. That changes how you make decisions. You stop building features nobody asked for, because you don't have the time. You stop chasing every interesting idea, because your hours are scarce. The constraint forces you to be ruthless about what actually moves a paying customer closer.

But the biggest reason I keep coming back to it: it forces you to be very smart with your time. When you only have a few hours a week, you can't grind your way to a result. You have to automate, systemise, and build leverage from day one. Every workflow has to run without you, because you're not there to run it. That's what makes the businesses I've built on the side actually scale — running them feels closer to checking my email inbox than working a second job. It's exactly what we teach in SoloStack: how to work smarter, not harder, by building businesses where the systems carry the load.

The five businesses, and how long each one took to build

Get Psyched — 12 months to build (2020)

The lesson: in 2020, the setup itself was the moat.

$500K revenue to date

I built Get Psyched because I was broke and good at explaining things. There was no grand vision. A psychology Honours student in Melbourne paid me to help her get through her thesis. Then another. Then another. I validated the problem in an afternoon of Facebook DMs. But the infrastructure to support a real business on top of it — website, booking flow, payment receipts, onboarding sequence, automated reminders, hiring tutors when I ran out of hours — took close to a year of nights and weekends. Half of that year was plumbing nobody could see. Five years later, it still pays me, including stretches where I barely touched it. But the year-long build is what kept most people from ever starting in the first place.

Online courses — 5 months to build (2023)

The lesson: same audience, new shape, half the build time.

$96K revenue to date

Once Get Psyched had hundreds of students paying for one-on-one tutoring, I packaged the lessons into self-paced online courses. Faster than Get Psyched, because the brand, the audience and the payment infrastructure were already there. But the course platform itself — video hosting, lesson delivery, progress tracking, the email sequencing that nudged people back in — was still five months of weekends. Same content, packaged once, sold many times. A service business pays you for your time. A digital product pays you for the work you already did.

TutorPro — 2 months to build (2024)

The lesson: AI started showing up on the build side, not just inside the product.

$15K revenue to date

By the time I built TutorPro, I was using AI to write the code, not just power the product. The build collapsed from months to weeks. The interesting thing about TutorPro isn't the product itself — it's that the customer list it launched onto took five years to accumulate. Content compounds. Trust compounds. The first 50 customers are the hardest. The next 5,000 are mostly word of mouth.

Augmented Designer — 2 weeks to build (2026)

The lesson: I stopped renting tools I used to outsource. Now I own everything.

$8K revenue since 2026

Augmented Designer is an online course portal with a custom CRM, payment gateways, lesson delivery, login and auth — the whole stack — built in a couple of weeks. The version of this I would have shipped five years ago would have meant Webflow for the site, Teachable for the course portal, HubSpot for the CRM, Stripe Checkout for payments, and Zapier to wire it all together. Six different bills. Three different dashboards. A developer on retainer for anything custom. Augmented Designer runs on one repo I own. The website is code. The CRM is a Postgres table. The course portal is a Vite app. Everything sits in one folder and shares one database. The first version was live in a week. The second iteration shipped the weekend after.

SoloStack — 1 week to build (2026)

The lesson: the playbook itself becomes the product.

SoloStack is the meta one. The same approach I've now used five times — service first, then digital product, then SaaS, then own-the-stack — is what we're packaging up to teach. The repo, the workflows, the AI agents, the CRM, the booking system, the proposal flow. Everything I wish I'd had on day one of Get Psyched. The first working version came together inside a single week, on top of the same Augmented Designer-style architecture — then shipped to other people as a workshop weekend.

The pattern across all five: the problem-validation work hasn't changed in six years. The relationship work hasn't changed. What's collapsed is the build window between "I have a working idea" and "I have a working business." Twelve months → five months → two months → two weeks → one week. That's the only thing about starting a business in 2026 that's genuinely different. And it's why we can now run the whole thing — setup, validation, build, first paying customer — inside a single workshop weekend. We've done it five times. The build window kept halving.

How each business fed the next

None of these businesses were built from scratch. Each one was built on the raw material the last one produced.

Each business compounded on the data the last one produced. The customer list grew. The content library grew. The same audience kept getting a tighter version of the answer. That's the part most "start a business" advice misses: the second business is supposed to be easier than the first, because the first one left you with assets.

Three things I now believe about starting from zero

The first thing I believe is that you don't need a plan. You need to be pointing in roughly the right direction. The details fill themselves in once you start moving. Every business I've started, I started because I took the first step. The plan caught up later.

The second thing I believe is that growth isn't linear. Year three of Get Psyched, I nearly walked away. The numbers were flat, I was exhausted, and the work felt thankless. I gave it one more reinvention. Year five it's still paying me. If you're in a flat month, that doesn't mean it's dead. It usually means the next version of the thing is about to arrive.

The third thing I believe is that the moment we're in is genuinely different. AI coding tools have collapsed the cost of building software from 'you need a team' to 'you need a laptop and an idea'. A solo founder in 2026 has access to the kind of build power that a venture-funded startup had in 2018. That's not hype. That's the reality of what Cursor and Claude Code do for someone willing to describe what they want in plain English.

The laptop was 2020's biggest unlock. Now AI is 2026's.

I keep coming back to that phrase. When I started Get Psyched in 2020, 'just my laptop' meant a Facebook page, a Stripe link, files going back and forth in DMs, and a lot of hours I spent doing the work by hand. The laptop was the tool, but most of the actual work was me.

In 2026 that phrase means something different. Now the laptop is a coding partner. I describe what I want, Claude Code writes the implementation, and the things I used to either buy as separate SaaS products or beg a developer friend to help with happen in an afternoon. A CRM that knows my customers. A landing page that actually converts. A booking system that talks to my calendar. An email sequence that follows up while I'm at my day job. All of it built in a folder on my laptop. All of it owned, not rented.

This is the part of the moment that I think a lot of people are still underestimating. The conversation about AI tends to live at the extremes. Either it's going to put everyone out of work, or it's a glorified autocomplete. The middle ground, the actually useful frame, is that AI has compressed the cost of building software from 'you need a team and a budget' to 'you need a laptop and a clear description of what you want'. The skill that scales now is not the ability to write code. It's the ability to look at output and say whether it's good. That's a judgment skill. That's where taste lives. And taste is the one thing that doesn't get automated away.

If you're someone with a real problem you've seen up close, a clear point of view on the people who have that problem, and a willingness to sit at a laptop for a few weekends, you're in the best moment to start something small that I've personally lived through.

$542/mo of SaaS in 2020. $13/mo in 2026.

Here's a comparison that still feels surreal to me. The stack of SaaS tools I would have needed to run Get Psyched in 2020 (a CRM, an email tool, a scheduling app, a payment processor, a website builder, an analytics tool, a place to send proposals, a sequence engine) would have run me somewhere between $400 and $600 per month. That's a real number. For a tutoring business doing a few thousand dollars a month, that overhead is the difference between profitable and not.

In 2026 the equivalent stack, built into one folder on my laptop, runs about $13 per month. Three dollars for Neon Postgres, ten for Resend email. Everything else is code I own. There is no per-seat pricing. There is no data silo tax. There is no Zapier middleman charging $50 a month to pretend that my email tool can read from my CRM. The whole company lives in one place, shares one database, and gets smarter every session.

That cost collapse is the companion story to the time collapse. Six years ago the question was "can you afford the tools to start a business this weekend?" Now the question is "can you describe what you want clearly enough that the AI can build it?" The first question kept a lot of people out. The second one is a skill anyone can learn — and learn fast.

The same skill works on your life, not just your business

The most underrated use of vibe coding isn't business. It's life infrastructure. The same skill that lets me ship Augmented Designer in two weeks lets me run multiple businesses on the side of a 9-to-5, with ADHD, without falling apart — because I've quietly vibe-coded a personal stack around myself. A morning brain dump that clears my head before email. A personal CRM that flags who I owe a reply to. An inbox triage agent. Each one a weekend. The collective effect is that "effortless multitasking" stops being a personality trait and starts being infrastructure. If you regularly feel paralysed or scattered, that's almost never an incompetence problem — it's the missing systems. I wrote the whole story up separately: paralysis isn't incompetence, you're just missing systems.

Why we can now teach this in a weekend

For most of the last decade, "start a business this weekend" was a marketing line, not a real claim. Even with the best intent and a clear idea, the build itself burned weeks before you were ever in front of a paying customer. Get Psyched took me a year. Online courses took me five months. TutorPro took me two. None of those timelines were a teaching problem — they were a build problem. The infrastructure between "I have an idea" and "someone can transact with me" was just genuinely slow to assemble.

That gap has closed. Augmented Designer went from idea to live customers in about two weeks. SoloStack itself went from idea to first working version in roughly a week. The reason the SoloStack workshop runs over a single weekend isn't that we've cut corners — it's that the build window has actually collapsed to fit inside one. We've watched it happen five times. The validation work hasn't changed. The selling work hasn't changed. What's gone is the unpaid setup gauntlet that filtered out almost everyone who tried.

If you've ever thought "I wish I knew how to start," the version of the answer I wish I'd had six years ago is now teachable in two days. That's SoloStack. To go deeper into any one piece, the next reads are the Cursor and Claude Code setup primer (about five minutes, even if you've never written a line of code) and the validation loop — the one thing AI didn't change, and the part you still have to do well.