We acquire and transform businesses where high-tech meets low-tech. Capturing the spread between analog labor and digitized intelligence.
We are not technology investors; we are efficiency maximalists. We believe the world is undergoing a Cognitive Industrial Revolution. Just as the steam engine decoupled physical force from human limitation, Artificial Intelligence decouples cognitive execution from human bandwidth.
We are currently living through a 20-year labor repricing event. The cost of intelligence is trending toward zero, yet the value of judgment and trust is trending toward infinity. Our fund exists to capture the spread between the old cost of doing business (analog labor) and the new cost (digitized intelligence).
We do not scale through capital. We scale through operational bandwidth. Our portfolio is deliberately constrained by the number of businesses we can personally transform.
Governing dynamics that do not change.
Money is merely a stored unit of time. Businesses that sell 'time back' to their customers have infinite demand; businesses that waste their owners' time have negative equity. We do not buy complexity; we buy the potential for speed.
We respect the Lindy Effect: The longer a non-perishable thing has survived, the longer it is likely to survive. We do not bet on new behaviors. We bet on ancient human needs (shelter, services, logistics, maintenance) fulfilled by new, invisible machinery. The most durable businesses look boring from the outside.
Outcomes are not distributed normally; they follow a Pareto distribution (80/20). In an AI world, the gap between the median and the best widens. A business enhanced by AI does not become 10% better; it becomes 10x more scalable. We seek opportunities where capped downside (real assets/cash flow) meets uncapped upside (margin expansion via automation).
In a technologically accelerated world, you must run fast just to stay in place. Stagnation is death. However, speed without direction is just friction. We deploy technology not to 'keep up,' but to sever the link between revenue growth and headcount growth.
Every moment we spend on a mediocre asset is a moment stolen from a great one. We move with conviction. We buy the past to build the future. We are the bridge between the analog foundation and the digital ceiling.
Understanding our nature.
Humans are evolutionarily wired to fear loss more than they desire gain. We buy businesses that solve painful, non-discretionary problems. We do not sell vitamins; we sell painkillers.
People do not want things; they want what other people want. Status drives behavior. However, in B2B and services, Trust is the ultimate status symbol. Trust is built in drops and lost in buckets. In an era of AI-generated noise, high-trust relationships are the ultimate scarcity.
No customer wakes up wanting a 'Language Model.' They want a clean house, a filed tax return, or a shipped package—faster, cheaper, and with less risk. AI is the plumbing, not the product. If the customer notices the AI, we have failed.
Customer behavior changes slowly, then all at once. We are currently in the accumulation phase of AI adoption. The 'phase transition' (mass adoption) occurs when the friction of using the new tech drops below the friction of the old habit. We invest before the transition becomes obvious to the herd.
How we evaluate and operate.
The highest ROI exists where high-tech meets 'low-tech' industries. We hunt in the unsexy valleys: logistics, HVAC, compliance, boring SaaS, and specialized services. These sectors suffer from labor shortages and aging owners (Demographics are Destiny). They are primed for an efficiency injection.
Complexity is a form of laziness. It is hard to make things simple. We ruthlessly simplify operations before we automate them. Automating a broken process just makes you break things faster.
In an age where code is commoditized, 'proprietary software' is rarely a moat. The durable moats in an AI world are:
We are not here to fire everyone. We are here to remove bottlenecks. AI substitutes drudgery, but it leverages judgment. A human + AI is not 1+1=2; it is 1+1=10. We restructure cognitive work like manufacturing was restructured in the 1980s—turning artisans into operators of high-leverage machinery.
What we believe is coming.
Think February 2020—a few people noticed a virus; three weeks later, the world changed. We are in that moment for cognitive labor. The builders of AI are not predicting; they are reporting what already happened to their own jobs. AI task completion time doubles every 4-7 months. The capability is here; deployment is the only bottleneck—and we are the deployment layer.
Software engineers watched AI go from 'helpful tool' to 'does my job better than I do' in 12 months. Every knowledge profession is next: law, finance, medicine, accounting, consulting, design. The builders say 1-5 years. We buy businesses where the human shifts from 'doer' to 'director'—the trust layer, not the labor layer.
The question is no longer 'How many hours did I work?' but 'How much machine work did I orchestrate?' The most productive person designs plans that machines run continuously. Leadership becomes scaling others' leverage, not your own hours. We restructure acquired businesses around this metric. Keep the GPUs full.
PhD-level engineers, accountants, marketers, strategists—available on demand, in your pocket, for $20/month. The bottleneck has shifted. It is no longer 'Can I find someone smart enough?' It is 'Can I define the problem clearly enough?' Judgment, taste, and direction are the new scarcities. Execution intelligence is abundant.
The new workflow: AI attempts the task; humans handle exceptions. Most businesses haven't seriously tried this—they use AI as 'Google with more words.' Due diligence now asks: 'What percentage of tasks has someone actually attempted to automate?' The gap between 'tried properly' and 'not tried' is massive and exploitable.
Most companies are still ignoring this. Most people use free-tier AI (18+ months behind frontier). This creates a brief window—measured in months, not years—where early adopters gain disproportionate advantage. We estimate 18-36 months before the majority catches up. Every acquisition made today benefits from a wider spread than those made in 2027.
Any work that happens through reading, writing, analyzing, and deciding via keyboard is exposed. What survives longer: physical presence, relationships built over years, licensed accountability, regulatory moats. Seniority alone does not protect you. Domain expertise alone does not protect you. 'My field is special' does not protect you.
Future systems will rely less on explicit prompts and more on inferred context. Don't build businesses around 'prompt engineering' as a service. Build around proprietary context: data, workflows, customer relationships. The business with the richest domain context will have the best AI performance in that domain.
AI is now contributing to developing its own successors. Development velocity is no longer constrained by human researcher bandwidth. The gap between model generations may compress further. We do not bet on specific timelines; we bet on direction. The direction is unambiguous: more capability, faster iterations, broader applicability.
Better information flow and fewer bottlenecks improve decision-making. Skip connections: information flows directly where needed. Distributed processing: decisions at the edge where context is richest. We flatten hierarchies, reduce bottlenecks, and build organizations that adapt as fast as the technology.
Past automation created adjacent jobs—the displaced found new roles the machines couldn't do. AI is different. It doesn't leave static gaps; it fills them dynamically. Each release closes the weaknesses of the last. The refuge you find today becomes automated tomorrow. We do not bet against this; we structure for a world where cognitive labor displacement is permanent and broad.
Rockefeller's wealth was ~2% of US GDP. That benchmark has already been exceeded—before most AI impact has materialized. If AI generates $3T in annual revenue at $30T valuation, personal fortunes in the trillions become plausible. Democracy is backstopped by mass economic necessity; if that leverage disappears, the social contract may break. We structure investments to distribute benefit broadly—not from altruism alone, but durability.
Every constraint is dissolving. 'I don't know how to code'—describe what you want; it appears. 'I can't afford the team'—$20/month for specialists. 'I need expensive experts'—best tutor in the world, infinite patience, 24/7. If you've wanted to build something but lacked skills or capital, that barrier is largely gone. The people who thrive are the curious and adaptable.
What we thought might take 20 years may take 10. What we thought might take 10 may take 5. Early adopters gain edge: 2024-2025. Majority adoption: 2025-2027. New equilibrium: 2027-2030. We accelerate acquisition pace while the spread is widest. Speed over perfection. The muscle of continuous adaptation.
We do not have the luxury of waiting. Asymmetric returns accrue to those who act while others debate. One hour a day experimenting with AI—non-negotiable. Move now; the spread is at maximum. Buy the boring; the most transformable businesses look analog. Transform with conviction; half-measures waste the window. Leadership must be fluent; you cannot delegate this.
The future is not coming. It is here. It just hasn't knocked on most doors yet. It's about to.
Alexi, Gabriele and Luca