By George Pu | Founder Reality Podcast | December 2025
I'm running two companies from Toronto with five people. No VC funding. No San Francisco office. And we're successfully competing with 50-person teams that have raised millions of dollars.
How? By ignoring everything Silicon Valley told me I needed and focusing on three things that actually matter in 2025: Capital, Code, and Audience.
Not the old definitions. New ones, transformed by AI and accessible to anyone reading this right now.
Everyone tells you the same story: big team, San Francisco office, venture capital. That might have been true before ChatGPT launched in November 2022. It's not true anymore.
Here's what actually matters.
Capital: It's Not About How Much You Have
When most people hear "capital," they think: raise $5 million from VCs, maybe get into YC for $500K.
That's wrong.
Capital isn't about how much money you have. It's about how efficiently you deploy it and how long you can keep it working.
The old way:
Raise millions. Hire 20 people. Get fancy office in SF or NYC. Burn $200-500K monthly. Hope you find product-market fit before you die.
The new way:
Keep costs brutally low. Focus on recurring revenue from day one. Stay alive long enough to build something real.
SimpleDirect: Built for $20K, Not $5M
We built SimpleDirect's initial product for under $20,000. Not $5 million. Not even $1 million. Twenty thousand dollars.
How?
No expensive office. For the first couple years, we had no office at all. I see companies in pre-revenue stage spending hundreds or thousands monthly on expensive offices. That's not the way.
Lean team. We started with two people. Scaled to 14 at our peak (realized it was chaos). Quickly scaled back down to two, now five. Compared to bloated startups, we save hundreds of thousands to millions of dollars yearly in salaries.
AI tools over new hires. We spend $8,000 per year on AI tools vs. $200-300K per person. Simple math.
Toronto, not San Francisco. Toronto is 40% cheaper than SF in rent and cost of goods. That's $100,000+ saved every year with our current setup.
But here's what matters more than all of that: capital liquidity.
Ask yourself three questions:
- Can you deploy your capital anywhere?
- Can you reposition your capital fast if something's not working?
- Does it compound without your time?
If you're spending time on a consulting business that can't run without you, you're failing the capital liquidity test.
I'm being honest: I'm still figuring out how my businesses can run without me. They'll need my involvement for at least another year. But we have 50+ months of runway because we're cautious about costs.
The takeaway: The more efficiently you deploy capital, the longer your runway and the more freedom you have to pivot when things don't work.
Efficiency is king.
Code: You Don't Need To Write It Yourself Anymore
"George, I don't know how to code" - I hear this constantly.
After AI became mainstream, anyone can code. Or more accurately: anyone can direct code to be written.
Code used to require four years of school or an expensive bootcamp. Used to require a technical co-founder to write thousands of lines. When I started as a university student, I believed I needed a technical co-founder.
Now I'm both CEO and CTO of my businesses.
I use AI tools like Cursor, Claude Code, ChatGPT, and GitHub Copilot to ask questions without judgment, understand syntax, write architecture, and plan user stories.
We're a team of five. Two are developers (not including me). Including me, three out of five can write code. That's a competitive advantage.
Here's the brutal truth: I haven't written production code in two and a half years. But I review code consistently. I understand what my developers are building. I know enough to guide the team and give input.
One of the biggest mistakes I see founders make: They don't understand their codebase and excuse it with "I'm not technical."
That excuse worked before AI. It doesn't work anymore. We're in an AI-first world.
My Stack (What Replaced Six Engineers)
At our peak, we had six engineers. Now we have two. Here's how:
Cursor: My primary development tool. Tell Cursor what to do, it generates code, I review and approve/disapprove. It's like having a junior developer who doesn't fully understand what they're doing yet. You need oversight, but it works.
Claude: Strategy, content, problem-solving, brainstorming. It's basically my co-founder for decision-making. I specifically told Claude to be extremely honest with me. It has been.
ChatRoot: Automates almost all customer support queries. You could also use Help Desk, Intercom, Help Scout - pick the cheapest one.
GitHub Copilot: Code assistance.
MCP Servers: Sounds complicated, but it just gives AI context about your business complexity.
The Director vs. Cameraman Analogy
You don't need to write code. You need to direct it.
Think of yourself as a film director, not a cameraman. The director doesn't operate the camera, but they know exactly what shot they want, how it fits the story, and whether the cameraman got it right.
That's you with AI tools.
I started with four co-founders. Now it's just me and one other co-founder. We automated traditional co-founder roles:
- CTO: Cursor, Claude, and me
- CPO (Chief Product Officer): Customer interviews, AI tools, and me
- Content creation: Claude amplifies my voice (99% me, Claude distributes)
- Customer support: ChatRoot automation
- Code execution: Two developers in India
- Strategic planning: ChatGPT, Claude, and my frameworks
Before you hire anyone or bring on another co-founder, ask yourself: Can AI do this? Can AI + a contractor solve this?
If yes, you just saved equity and complexity.
Audience: The Most Underrated Lever
When people hear "audience," they think: "I need a million followers like Logan Paul or Mr. Beast."
Wrong.
Audience means a group of people who trust you enough to listen when you have something to say. This could be 200 people. Could be 200,000. Could be 2 million.
Size matters less than trust and reputation.
Why Distribution Without Permission Changes Everything
Traditional founder without audience:
If you're finding this useful, I send essays like this 2-3x per week.
·No spam
- Cold outreach on LinkedIn: 100 messages → 2-5% response rate
- Conferences: $10-50K for a booth + 200 business cards
- Paid ads: Google, Facebook, LinkedIn - expensive and burns runway
- Sales team: Expensive, and they don't know your product like you do
Founder with audience:
- Post about your business → immediate reach
- Zero ad spend, compounding distribution
- Direct access to people who already trust you
I've spent the past two years building to 30,000 Twitter followers. I post every day consistently. Two months ago, I started a newsletter to own the relationship (not just rely on social platforms).
Last week, one post got 150,000 impressions.
Compare that to a founder without presence: 200 LinkedIn messages per week, 2-10 responses, maybe 1-2 qualified leads. Then do it all again next week.
With audience, I don't need:
- TechCrunch coverage
- PR agency ($5,000/month - almost paid this)
- YC validation
- Conference booth access
I already have people opted-in to hear what I have to say. That's the power.
Trust > Follower Count
I've seen people with 200,000 followers get a few likes per post. They have following but no trust.
If you post memes, you can hit 500,000 followers easily. But the value to you is almost zero because it doesn't translate to anything except maybe selling ads.
If you build a personal brand, talk about your business, focus on a specific niche - even 500 followers who trust you and care about your brand matter more than 500,000 who don't.
When To Start (Hint: Before You Launch)
The biggest mistake founders make: thinking you build an audience AFTER you launch something.
Wrong. Start before you launch.
We don't follow brands. We follow people. That's the power of being a founder - you have a story. Everyone does. Childhood story, how you grew up, how you navigated difficulties.
It took courage for me to start sharing my story. It doesn't come easy. But you have to do it before you launch.
I started on Twitter in 2021 but didn't post actively until 2023. Two years to get to 30,000 followers. First few months? No likes, no impressions. Just shouting into the void.
But I kept going. One post per day minimum.
Rented vs. Owned Audiences
Rented platforms: LinkedIn, Twitter/X, YouTube
You don't control distribution. Algorithm decides who sees your content.
Owned platforms: Newsletter (emails), Podcast
You own direct distribution to your audience.
Start with rented channels - they give you the most initial boost. Once you have enough people and confidence, prioritize owned audiences. That way, you're not dependent on algorithm changes.
What Doesn't Matter Anymore
Headcount: At our peak, 14 people for a company making $30-40K/month. Insane. Eventually had to let everyone go.
With 14 people, there are 91 possible communication paths. With five people, there are 10. Way simpler.
The bigger your team, the more time coordinating. With five people, I know exactly who to talk to. Team knows who to reach out to. No bureaucracy.
Geography: I'm building in Toronto - not a tech hub, not a SaaS hub. But I'm finding success here because of lower cost of living and my support network.
I still go to SF and NYC for networking, but maybe a few days per year total. I can build from Bali, Beijing, Tokyo, Argentina, Colombia, anywhere in Europe with lower cost of living.
We use Zoom and Google Meet for networking. You don't need to meet in person constantly anymore.
Toronto being 40% cheaper than SF saves us $100,000+ yearly. That's not nothing.
The Three C Audit
Run this on your business today:
Capital:
- How many months of runway do you have?
- Can you cut costs by 30-50% without killing the business?
- Are you optimizing for efficient deployment or just spending what you raised?
Code:
- Do you understand your tech stack even if you can't write production code?
- Can AI + contractors replace the next hire you're considering?
- Are you directing code or trying to write everything yourself?
Audience:
- Can you reach people in 48 hours without paid ads?
- How many people trust you enough to listen when you launch something?
- Are you building audience before or after launch?
Fix what's broken. Double down on what's working.
From Machine-Platform-Crowd to Capital-Code-Audience
Years ago, I read "Machine, Platform, Crowd" - it was brilliant for 2015-2017. The idea: you need machines (computers/servers), platforms (marketplaces/SaaS), and crowd (audiences).
That model is outdated.
In 2025, after ChatGPT changed everything, the new model is Capital-Code-Audience. Three C's.
One person with capital that's liquid and efficient, code that multiplies output through AI, and an audience that trusts them can outcompete a 50-person team in a fancy San Francisco office.
My businesses are proof. Five people. Two companies. $2.35 million in revenue over three years. Toronto, not SF. No VC funding.
Not because I'm smarter. Because I focused on the three things that actually matter and ignored Silicon Valley's playbook.
You can do the same.
What I'm building: SimpleDirect and ANC with the Three C Framework
Daily insights: @TheGeorgePu on Twitter
Newsletter: newsletter.founderreality.com
George Pu builds AI-powered businesses at SimpleDirect and ANC. Follow along for unfiltered founder insights at @TheGeorgePu.