At Fractal, we built companies from scratch. The studio incubated around 150 companies in just under three years. I launched 32 of them across verticals — each one starting from a blank page, a thesis, and a few conversations that turned into something worth building.

What made Fractal work was the machine we built to power our ideas.

Most venture studios fail because they treat company building like art — inspired founders with unique insights creating one-of-a-kind businesses. Fractal succeeded because we treated it like science. We built a research engine designed to find billion-dollar opportunities in "boring" markets, developed repeatable playbooks for taking concepts to revenue, and created organizational infrastructure that let us scale from launching one company per month to ten.

The result was one of the most productive venture studios in history.

From 2020 to 2023, Fractal grew from a handful of people to ~120 employees across research, talent, and portfolio support. We raised over $650M from top LPs and incubated a portfolio of ~150 vertical software companies that serve the bedrock economy. Fractal proved that company building could be systematized, that "boring" markets contained enormous opportunities, and that the right infrastructure could dramatically accelerate the zero-to-one journey.

This is the story of how we built it.

Fractal's thesis was counterintuitive: the best startup opportunities exist in industries nobody talks about. Not sexy consumer apps or cutting-edge AI. Logistics software. HVAC maintenance. Medical billing. Commercial lending. Industries with sneaky big TAMs, entrenched inefficiencies, and digital transformation happening at a snail's pace.

Venture capital had spent decades chasing "massive addressable markets" in consumer tech and horizontal SaaS. The result was brutal competition, inflated valuations, and winner-take-all dynamics where 90% of startups failed. Meanwhile, entire sectors of the economy — collectively worth trillions — remained underserved by modern software.

The founding team at Fractal recognized this was exactly backward.

Industries that VCs ignored because they seemed "too small" or "not sexy enough" were actually perfect for company building. They had real customers with real budgets solving real problems. They lacked modern software infrastructure. They were fragmented across thousands of SMBs, each desperate for better tools but underserved by existing vendors.

Most importantly, they were defensible. Building software for HVAC contractors requires deep domain expertise. You need to understand their workflows, their pain points, and their industry-specific terminology. Generic horizontal tools don't work. This creates natural moats — once you build authentic domain expertise and customer relationships, competitors can't easily displace you.

The insight was profound: create a systematized approach to finding these opportunities, validating them quickly, and building companies around them at scale.

If done with discipline and proper infrastructure, the math was compelling. You could build hundreds of defensible, cash-generating businesses in markets large enough to support venture outcomes but small enough that you wouldn't face hypergrowth competitors burning capital to win at any cost.

But first, you need to find the right opportunities. With hundreds of potential industries to explore and thousands of discrete problem spaces within each, we needed a systematic way to identify where to focus.

We built a research function designed to map the entire landscape of "boring" markets and surface the highest-potential opportunities. The work was equal parts research, anthropology, and pattern recognition. We mapped value chains, analyzed incumbents, studied regulatory environments, and identified where time, money, or information was being wasted.

The research process followed a structured framework. First, we'd identify industries with specific characteristics: sizable TAM, high fragmentation, outdated technology infrastructure, and willingness to pay for better solutions. We looked for industries where software penetration was low but business complexity was high — a sign that existing tools weren't meeting needs.

Then, we'd go deep on value chain analysis. Where does money flow in this industry? Who are the key stakeholders? What workflows are most painful? Where is information getting lost or time being wasted? This required actual research — reading industry publications, analyzing financial data, studying incumbent solutions, and mapping the ecosystem of vendors, suppliers, customers, and regulators.

Next, we'd validate through primary research. This meant talking directly to business owners — thousands of them over the years. Restaurants, logistics firms, clinics, construction companies, property managers, manufacturers. The best insights didn't come from secondary data. They came from listening.

When someone said "I hate this part of my job," that was usually the beginning of something. When they described workarounds they'd built in Excel or processes that required manual data entry across multiple systems, we knew there was an opportunity. When they told us they'd pay for a solution that worked, we knew there was a business.

The research engine produced a pipeline of thousands of potential ideas. Industries to explore. Problems to solve. Business models to test. The output wasn't just concepts — it was structured theses with market sizing, competitive analysis, customer segmentation, and go-to-market hypotheses.

Finding the signal inside that noise was the real craft.

Most markets look promising from afar. Few can support a venture-scale business. Our job was to get close enough to see the truth fast enough that we weren't wasting months on dead ends.

The incubation process followed a tight, repeatable loop designed to de-risk ideas as quickly as possible.

Stage one was problem validation. Before building the MVP, we'd spend a few weeks validating that the problem was real, acute, and worth solving. This meant more customer conversations, at times hundreds of in-depth interviews with target customers. We'd present the problem as we understood it and see if it resonated. We'd ask about their current solutions and whether they'd tried alternatives. We'd probe on budget, urgency, and decision-making authority.

The goal was to kill ideas fast. If we couldn't find enough people who considered this a top-three problem, we'd pivot or kill the concept. If existing solutions "worked well," we'd move on. If the problem was real but the willingness to pay was weak, we'd reconsider the business model.

Stage two was solution prototyping. Once we validated the problem, we'd build the early product within a month or two. Not a full product or production-grade infrastructure. Just enough to demonstrate that we understood the problem and could deliver value.

This is where most studios fail. They build too much too soon. They spend six months perfecting a product before talking to customers. They over-engineer solutions for problems they don't fully understand.

We built fast and iterated faster. Mock-ups and prototypes that could demonstrate the core workflow. Landing pages that explained the value proposition. Sometimes, even manual processes where we'd deliver the service ourselves before building software, just to prove customers would pay for the outcome.

Stage three was revenue validation. In the research process, we would sell the prototype and vision to ~20 pilot customers. We looked for real commitment — not just interest, but willingness to switch from their current solution, invest time in onboarding, and integrate it into their daily workflow. This was crucial. Customers will tell you anything in an interview. They'll only change their behavior for things they actually value.

If pilots were successful, we'd move to stage four: building the founding team and taking it from zero to one. We'd recruit the founders (sometimes from within Fractal, sometimes from our network, sometimes from the industry itself), build a small team around them, and give them the resources to scale from pilots to real revenue.

The entire loop from initial research to founding team typically took 3-4 months. Some companies moved faster. Others required more iteration. But the framework was consistent: validate the problem, prototype a solution, prove customers will pay, then build the team to scale it.

Early on, Fractal could only launch a handful of companies at a time. The process was too manual, too dependent on a few people doing everything. Each incubation required deep involvement from the founding team. Knowledge lived in people's heads rather than in documented systems. New team members took months to ramp up because there was no playbook to follow.

This was the constraint on growth.

If we wanted to go from launching 5 companies per year to 50, we needed to systematize everything. Research frameworks, GTM blueprints, product guides, growth checklists, hiring rubrics, financial models. We needed to codify what had worked in a handful of companies and make it repeatable for the next hundred.

So we built the systems to scale it.

On the research side, we developed standardized frameworks for industry analysis, customer segmentation, and competitive assessment. We created templates for research memos that forced structured thinking: What's the TAM? Who are the incumbents? What's the value chain? Where's the inefficiency? What's our go-to-market motion? What are the risks?

These frameworks did two things. First, they ensured consistency — every research project covered the same ground with the same rigor. Second, they enabled knowledge transfer. A researcher could hand off a project to someone else, and they'd immediately understand the state of the analysis.

On the incubation side, we built playbooks for each stage of company building. How to run effective customer discovery. How to structure pilot programs. How to price early offerings. How to recruit founding teams. How to structure equity and compensation. How to think about product roadmaps and prioritization.

We documented everything. Not rigid processes that killed creativity, but flexible frameworks that gave people structure to move with velocity. The goal was to make the implicit explicit — take the tribal knowledge that lived in a few people's heads and make it accessible to the entire organization.

We also built operational infrastructure. Financial models that projected burn, revenue, and headcount for each portfolio company. CRM systems to track customer conversations and deal flow. Project management tools to coordinate across research, product, and GTM. Dashboards that gave leadership real-time visibility into every incubation's progress and post-launch performance.

That's what allowed Fractal to evolve from an experiment into a machine.

Within a year, we went from incubating one company per month to ten.

That required hiring and training a team capable of executing the playbook, building feedback loops between research and portfolio so learnings from each company fed back into the system, and creating enough organizational capacity that we could run multiple parallel workstreams without creating bottlenecks.

The learnings from each company compounded knowledge with every new launch. We learned which industries had the best economics. Which go-to-market motions worked for SMBs versus enterprises. Which pricing models drove adoption. Which types of founders succeeded. All of that fed back into how we evaluated new opportunities and structured new incubations.

When you build 32 companies, you start to see patterns everywhere.

You learn what good looks like before the numbers show it. You can sense when a founder has clarity about their customer and their value proposition. When a product resonates. When a story lands. You develop intuition about team dynamics, market timing, and competitive positioning that's hard to articulate but easy to recognize.

You also develop humility.

For every company that took off, there were others that didn't. Some failed because the market wasn't as big as we thought. Others because we couldn't find the right founding team. Others because timing was wrong or competition moved faster than expected. Building at this scale teaches you that execution matters more than theory, that team quality trumps everything else, and that even great ideas can fail with poor execution.

The best founders I worked with had two traits in common: curiosity and focus.

They listened more than they talked. They invested the time to understand their customers before building anything meaningful. They asked why repeatedly until they understood the root cause, not just the surface symptom. They were honest about what was working and what wasn't, willing to pivot when data contradicted their assumptions.

And they had focus. They didn't chase trends or pivot every time a new opportunity appeared. They picked a problem, went deep on it, and solved it better than anyone else. That focus created defensibility — deep domain expertise that competitors couldn't replicate quickly.

The boring problems were usually where the value was hiding. Everyone wants to build the sexy consumer app or the cutting-edge AI platform. Few people want to build software for commercial lending or HVAC maintenance. But that's exactly where opportunity exists — in the markets everyone else ignores.

Fractal taught me several lessons that apply far beyond venture studios.

Research only matters if it leads to action. We could have spent years mapping industries and analyzing markets. Instead, we built bias toward shipping — getting something in front of customers fast enough to learn whether we were right. The research was valuable because it informed action, not because it was comprehensive.

Process beats intuition when you're building at scale. A few brilliant people can build great companies through instinct and pattern matching. But if you want to build dozens of companies simultaneously, you need systems. Frameworks that capture institutional knowledge. Playbooks that enable new team members to execute without reinventing everything. Processes that ensure consistent quality across a large portfolio.

Structure isn't bureaucracy — it's what frees people up to think and create. The research frameworks, incubation playbooks, and operational infrastructure we built didn't slow people down. They eliminated ambiguity and repetitive work, giving people more time to focus on the problems that actually required creative thinking.

The best opportunities are often the least obvious. VCs chase massive horizontal or consumer markets because they're visible and exciting. But the fragmented, boring industries generate better risk-adjusted returns. Less competition. More defensibility. Customers who will actually pay for solutions. These aren't markets you can pattern-match from Sand Hill Road — you have to do the work to find them.

Company building is a team sport. We never launched companies with solo founders. The best outcomes came when we assembled full teams early — co-founder pairs + product + GTM + domain experts. That collective intelligence moved faster and made better decisions than any individual could alone.

Fractal was a learning engine, and for me, it was the best possible training ground for understanding how companies actually get built at scale.

What we proved was that company building could be systematized without losing the creativity and intuition that make startups work. Those boring markets contain enormous opportunities if you're willing to do the research. The right infrastructure can dramatically accelerate the zero-to-one journey. That focus and execution matter more than perfect ideas.

The venture studio model isn't for everyone. It requires significant upfront capital, the ability to attract entrepreneurial talent without the full founder equity package, and the operational sophistication to manage a complex portfolio. Most venture studios fail because they can't solve one or more of those challenges.

Fractal succeeded because we built the full stack — research engine, incubation playbook, operational infrastructure, and talent pipeline. We didn't just generate ideas or write checks. We did the hard work of validating markets, building products, acquiring customers, and assembling teams. We created the systems that made all of that repeatable.

The 32 companies I launched weren't just learning experiences. They were proof that the model worked. That with the right approach, you could find opportunities others missed, validate them quickly, and build defensible businesses around them — not once, but over and over again.

What I took away wasn't just operational muscle. It was pattern recognition — the ability to see markets, teams, and timing clearly. The ability to know when to push harder and when to pivot. The belief that even in a world obsessed with speed, patience and precision still win.

And most importantly, the conviction that the best opportunities are often hiding in plain sight, waiting for someone willing to do the work to find them.