The Flywheel Effect: How Our Ventures Feed Each Other_
The Flywheel Effect: How Our Ventures Feed Each Other
Jim Collins has this image that I can't shake. A massive, heavy flywheel — thousands of pounds of metal mounted on an axle. You push it. Nothing happens. You push again. It moves an inch. You keep pushing. One revolution. Then two. Then — at some point you can't precisely identify — the wheel's own momentum starts carrying it forward, and each push adds to a force that's now nearly unstoppable.
The flywheel doesn't care about your strategy deck. It doesn't care about your vision statement. It only responds to consistent, directional force applied over time.
When I designed AI Gens's structure, I didn't start with a spreadsheet of projected returns or a matrix of market opportunities. I started with a question: how do I build a system where every venture we create makes every other venture stronger?
Not a portfolio. A flywheel.
The Traditional Fund Problem
Here's what most venture capital looks like from the inside: you raise a fund, deploy capital into 30-50 companies over three years, provide board seats and introductions, and wait. Each company is its own island. They share a brand name on their cap table and access to the GP's network, but they don't share technology, infrastructure, customers, or hard-won operational learnings.
When Portfolio Company A discovers that a particular go-to-market strategy works brilliantly for developer tools, that insight lives in the partner's head and maybe in a newsletter. It doesn't become infrastructure that Portfolio Company B can build on. When Company C builds a sophisticated data pipeline, Company D starts from scratch. The intellectual capital of the portfolio — the accumulated wisdom from building, failing, iterating, and succeeding — dissipates instead of compounds.
This is a structural problem, not a people problem. The traditional fund model is designed for financial diversification, not operational synergy. The incentive structure (management fees + carry across a large portfolio) rewards breadth, not depth. The legal structure (each company as an independent entity with its own board) prevents the kind of deep operational integration where real synergies live.
The AI Gens Flywheel
AI Gens's flywheel has six segments, each feeding the next:
Segment 1: Consulting Revenue. Moonxi, our consulting studio, generates revenue by delivering platform engineering and AI services to enterprise clients. This is the "safe" side of our barbell — proven demand, recurring relationships, predictable cash flow. It's not glamorous, but it's the gravitational force that starts the wheel turning.
Segment 2: Product Development. Consulting revenue funds the development of our product ventures — Apollo, HOST360, Mustard. But it funds them in a specific way: not as abstract R&D projects disconnected from reality, but as solutions to problems we encounter daily in consulting engagements. Apollo was born because we needed a delivery control plane. HOST360 emerged from patterns we saw across multiple hospitality clients. This isn't incubation theater. It's necessity-driven product development.
Segment 3: IP Creation. As product ventures mature, they create intellectual property — not just code, but frameworks, methodologies, governance models, and operational patterns. Apollo's agent governance framework. HOST360's multi-property management architecture. Mustard's brand engineering pipeline. Each becomes a reusable asset.
Segment 4: Consulting Enhancement. This IP flows back into consulting. When Moonxi pitches a new engagement, we're not selling generic engineering hours. We're selling battle-tested frameworks backed by products we use ourselves. "We built a delivery control plane that governs 203 agent tools — we can implement something similar for your organization." The consulting becomes more valuable because the products exist, and the products become more credible because the consulting proves them.
Segment 5: Margin Expansion. Better IP in consulting means higher value per engagement, shorter delivery cycles, and lower cost of delivery. Margins expand. Where a generic consultancy might achieve 15-20% margins on engineering services, a consultancy armed with proprietary frameworks and tooling can push toward 30-40%. Those expanded margins create surplus capital.
Segment 6: New Venture Funding. Surplus capital funds the next venture. And critically, each new venture benefits from all the IP, infrastructure, and learnings from the previous ones. HOST360 doesn't need to build its own deployment infrastructure — it uses the same platform Moonxi runs on. Mustard doesn't need to create its own agent governance — it inherits Apollo's framework.
Then the cycle repeats. Each rotation faster than the last.
How Intellectual Capital Compounds
In traditional finance, compound interest is well understood: returns generate more returns, and the curve goes exponential. What's less discussed is that intellectual capital compounds the same way — but only if you build the infrastructure for it to flow.
Here's a concrete example. In 2025, while building Apollo, we developed a pattern for governing agent access to sensitive client data. The pattern involved policy envelopes — metadata wrappers that define what an agent can do, under what conditions, with what audit trail. We implemented it for Apollo's 203 tools.
Six months later, when we started building HOST360's guest experience agent, we needed to govern agent access to guest personal information — different domain, same structural problem. Instead of redesigning from scratch, we ported the governance framework. What took three months for Apollo took two weeks for HOST360.
Then Mustard needed agent governance for brand-sensitive content generation. Same framework, new policies. One week.
That's intellectual capital compounding. Three months became two weeks became one week. The knowledge didn't just transfer — it accelerated.
In a traditional fund, these three ventures would be independent portfolio companies. Each would hire its own architects, make its own governance mistakes, and spend its own months reinventing patterns that already exist somewhere in the portfolio. The IP would stay siloed. The learning curve would restart each time.
The Hedgehog Concept
Collins introduced another framework that I find even more useful than the flywheel for making strategic decisions: the Hedgehog Concept. It's the intersection of three circles:
What are you deeply passionate about? For AI Gens, this is building infrastructure companies. Not apps. Not features. Not marketing tech or social platforms. Infrastructure — systems that other things are built on top of. I've been obsessed with this since I was an engineer building deployment pipelines. The question that animates me hasn't changed in fifteen years: how do you build something that becomes foundational?
What can you be best in the world at? Platform engineering combined with AI. This is specific enough to be meaningful and broad enough to be applicable. We're not claiming to be the best at machine learning research or the best at enterprise sales. We're claiming a narrow intersection: building platforms that leverage AI to become infrastructure. The team we've assembled — platform engineers, AI practitioners, DevOps specialists — is configured for this exact intersection.
What drives your economic engine? Two fuel sources: consulting revenue (steady, predictable, provides operational cash flow) and venture equity (volatile, long-term, provides asymmetric upside). The combination is unusual. Most consultancies don't build ventures. Most venture firms don't run consultancies. But the dual engine means we're not dependent on fundraising cycles for survival, and we're not limited to consulting margins for growth.
The Hedgehog Concept acts as a filter. When an opportunity comes along — a potential client engagement, a venture idea, a partnership — we ask: does it sit at the intersection of all three circles? If it requires passion we don't have, capability we can't develop, or doesn't feed the economic engine, we pass. Even if it looks profitable in isolation.
This is how we avoid the trap of the unfocused studio that incubates whatever seems exciting this quarter. Every venture must be an infrastructure play (passion), must leverage platform engineering + AI (capability), and must feed either consulting revenue or venture equity (economic engine). No exceptions.
Why Flywheels Are Brutal to Start
I want to be honest about something that Collins acknowledges but that most people who cite the flywheel metaphor conveniently skip: the early rotations are miserable.
For the first two years of AI Gens, the flywheel felt like a metaphor for futility. Consulting revenue was modest. Product development was slow because we were funding it from operating cash flow, not venture capital. The IP we were creating was raw — useful to us, but not yet refined enough to differentiate our consulting. Margins were tight.
There were months where I questioned whether the studio model was just an elaborate way to run a consulting shop with delusions of venture grandeur. The spreadsheet said "flywheel." The lived experience said "treadmill."
What changed wasn't any single breakthrough. It was accumulation. Enough consulting engagements that we understood patterns across industries. Enough product iterations that Apollo actually worked reliably. Enough IP that clients started saying, "Wait, you built that? Can we use it?"
The flywheel didn't suddenly start spinning. It incrementally stopped resisting.
Cross-Portfolio Learnings
One of the least discussed advantages of the studio model is what I call cross-portfolio pollination — insights from one venture that transform another in ways you couldn't have predicted.
Building HOST360 taught us something about Apollo. Hospitality operators think in terms of guest journeys — arrival, stay, departure — not in terms of engineering sprints. When we mapped HOST360's journey model back to Apollo, we realized that consulting delivery also has journeys: engagement start, delivery execution, handoff. Apollo's interface was organized around engineering concepts (sprints, deployments, costs). Reframing it around delivery journeys made it more intuitive for non-technical stakeholders.
That insight didn't come from user research or design thinking workshops. It came from building in an adjacent domain and noticing structural similarities.
Similarly, Mustard's brand engineering work — using AI to maintain brand consistency across touchpoints — taught us something about agent governance. Brand guidelines are essentially policies. "Use this tone of voice, never use these words, always include this disclaimer." The enforcement mechanism is structurally identical to Apollo's tool governance: a policy envelope that constrains agent behavior within defined boundaries.
These cross-pollinations are the flywheel's hidden power. They don't appear in financial projections or pitch decks. They emerge from the practice of building multiple things simultaneously and paying attention to the patterns that connect them.
Early Rotations
AI Gens's flywheel is in its early rotations. I want to be clear about that. We're not yet in the exponential phase where momentum carries itself. We're still pushing.
But the wheel is moving. Each rotation is noticeably faster than the last. Consulting engagements close faster because the IP is real. Product development is faster because the infrastructure is shared. New ventures start further along because they inherit frameworks instead of building from scratch.
The brutal part is behind us. The exponential part is ahead. The part we're in now — where you can feel the momentum building but haven't yet achieved escape velocity — is, in some ways, the most interesting. Because this is where discipline matters most. This is where the temptation to take shortcuts, to pursue opportunities outside the Hedgehog Concept, to optimize for short-term revenue at the cost of long-term compounding — this is where that temptation is strongest.
The flywheel demands consistency. Not inspiration. Not brilliance. Consistency.
We push. The wheel turns. We push again.