He Built 10 AI Companies in 12 Months - With Teams of Five
with Brennan, Infinity Constellation
He Built 10 AI Companies in 12 Months - With Teams of Five
Show Notes
The traditional startup playbook says you pick one idea, one market, and go deep for years before you see whether it works. Brennan threw that playbook out.
In twelve months, his company Infinity Constellation launched ten AI-native service businesses - each with a team of fewer than five people, each paired with a repeat founder, and each targeting a specific slice of the $6 trillion global service economy that has never been seriously disrupted by software. Multiple companies reached $1 million in annual recurring revenue in their first year. Total capital raised: $17 million.
This episode is about what becomes possible when the cost of building falls so fast that the old constraints of capital, headcount, and time no longer determine who gets to build companies - and what the right model looks like when that shift is permanent.
The AI Holding Company Thesis
Infinity Constellation's model is deliberately inspired by Berkshire Hathaway - a holding company that creates and operates a portfolio of businesses under shared infrastructure rather than picking one bet and scaling it. But where Berkshire acquires mature businesses, Infinity Constellation builds AI-native service companies from scratch, pairs them with experienced founders, and operates them with small teams that would have been impossible before AI.
The thesis rests on a structural claim: for the first time in history, you can deliver a service outcome at enterprise scale with a team of three to five people. Not because the team is more talented, but because AI has collapsed the labor cost of execution. A marketplace that took two to three years and ten engineers to build pre-AI can now be built in twenty-eight days with one engineer. That compression changes what is economically rational to attempt.
The portfolio spans recruitment (Zero Hiring), design (an IDEO challenger), executive assistance (Everest), philosophy and classics education (Library of Alexandria), regulatory permitting (Labyrinth), IT services (Sanctum), and business development (Ascendancy). Each targets a service category with historically high labor costs and low software penetration.
Frameworks from This Episode
These frameworks have been added to the AI for Founders Frameworks Library. Filter by Brennan (Infinity Constellation) to find them.
The AI Holding Company Model
Build a portfolio of AI-native service businesses under shared infrastructure rather than scaling one product. The holding company captures diversified upside across multiple category disruptions while sharing GTM, talent, and capital allocation expertise. Each company stays small by design - five people is a feature, not a constraint.
Outcome-as-a-Service
Sell the result the customer needs - the hired candidate, the delivered design system, the filed permit - not software access. AI makes this economically viable at scale for the first time because the marginal cost of executing the outcome falls dramatically. The customer pays for what they actually wanted; the company captures the efficiency gains from AI execution.
Good Headcount vs. Bad Headcount
Good headcount augments human judgment with AI leverage - one expert doing the work of ten, with AI handling execution. Bad headcount is the elevator operator model: a human performing a task that exists only because the technology to automate it had not yet arrived. The test: if the job description would have been written the same way in 2019, it might be an elevator operator role.
Tools from This Episode
The world's first AI holding company. Builds AI-native service businesses from scratch and pairs them with experienced founders.
This Week's Experiment
Write Your Company Memo: From Tool to Outcome
Take your current product or service and rewrite the value proposition as an outcome, not a tool. If you sell project management software, the outcome is "projects shipped on time." If you sell recruiting software, the outcome is "the right hire in your seat in 30 days." Write a one-page memo making the case for how you could charge for the outcome itself - and what AI infrastructure you would need to actually deliver it. The exercise clarifies where your AI leverage is real and where you are still selling a vitamin.
The Tech Moat Collapse
One of Brennan's central claims is that the technology moat - the competitive advantage that came from having engineering resources large enough to build complex software - has collapsed. A marketplace that took two to three years and ten engineers can now be built in twenty-eight days with one engineer. That is not an incremental improvement; it is a category destruction.
The implication: service businesses that believed they were protected from software disruption by the complexity of their workflows no longer are. Regulatory permitting, recruitment, IT services, professional design - these are categories where the software to automate the work now exists and can be deployed by a small team. The question is not whether disruption is coming but which founder gets there first.
Founder-Led Distribution and the Death of Cold Outbound
Brennan's GTM thesis is blunt: cold calling and cold email are largely dead. The founders who are winning are using content, podcasts, events like Abundance 360, and community to generate inbound demand before the first sales conversation. He draws the Palantir parallel: Alex Karp as the face of the company, going on every stage that would have him. That founder visibility does not just generate leads - it creates a category association in the buyer's mind that is hard for a faceless sales team to replicate.
The structural advantage of founder-led sales scales further than most founders assume. Brennan's observation: the best salespeople he has seen are the founders themselves, because they understand the problem at a level no hired rep can replicate in the early years. Hiring a sales team before you have founder-led traction is often an attempt to scale a machine you have not yet proved can run.
Q&A
What is Infinity Constellation and what makes it different from a traditional startup?
Infinity Constellation is a holding company that builds AI-native service businesses from scratch rather than acquiring existing ones. Unlike a traditional startup that bets on one product in one market, the model creates a diversified portfolio of companies - each pairing a repeat founder with an applied AI specialist - targeting high-labor, low-software-penetration service categories. The model is deliberately inspired by Berkshire Hathaway but applied to creation rather than acquisition. Each company stays small by design: teams of fewer than five people deliver outcomes that previously required twenty to fifty.
How did ten companies launch in twelve months?
The holding company model uses shared infrastructure across the portfolio: shared capital allocation, shared GTM expertise, shared talent recruitment, and cross-portfolio learnings on what works. Each company does not start from zero on operational questions - they inherit lessons the holding company has already solved. The founder pair model accelerates speed: pairing a repeat founder with an applied AI person compresses time from idea to first revenue significantly.
What does outcome-as-a-service actually mean in practice?
Rather than selling software access, you sell the deliverable the customer actually wanted. Zero Hiring sells placed candidates, not recruiter software. A design agency sells a delivered design system, not design tool seats. The economics work because AI reduces the marginal cost of executing the outcome dramatically - a small team can now do what previously required a large team. The customer pays a premium for the outcome, and the service provider captures the efficiency gain from AI execution.
What is the elevator operator principle?
Before automatic elevators, buildings employed human operators to run elevator cars. When automatic elevators arrived, those jobs disappeared - not because the workers were bad at their jobs but because the technology made the role redundant. Brennan applies this to headcount decisions today: any job that could be described the same way in 2019 as today is a candidate for being an elevator operator role. Bad headcount is people doing jobs AI can do. Good headcount is people doing judgment work that AI assists but cannot replace - the specialist who is ten times more effective because of AI tools, not the person who exists because the automation has not yet been built.
How does Infinity Constellation choose which service categories to enter?
The filter is the intersection of three conditions: high labor cost (service businesses where human execution accounts for the majority of delivery cost), low existing software penetration (categories too complex or too workflow-specific for standard SaaS), and large existing market. Regulatory permitting, recruitment, IT services, and design are all mature industries that have not been disrupted by software not because the problems were not valuable, but because the software to automate the workflows was too expensive to build until now.
What is multiplayer entrepreneurship?
Brennan's frame for the AI holding company model: instead of one founder solving one problem alone, you run multiple companies in parallel under shared infrastructure, with different founder-operator pairs driving each one. The holding company layer handles capital, recruiting, and cross-portfolio learning while each company's founding team drives the business independently. Multiple teams pursue different objectives within a shared world, with coordination at the infrastructure layer but genuine autonomy at the execution layer.