Most startups don't die because they ran out of money. That's the symptom on seventy percent of the death certificates, not the disease. In this episode we rank the real causes by how often they actually kill, counting down to the true number one. Broken unit economics (19%) is the weakest, survivable cause, almost every fast-growing company passes through it. Founder conflict (23%) gets named constantly but it's a downstream symptom, a moderatable risk. Bad timing (29%) can be forced with enough money, but it's brutal for a small startup, see our Better Place teardown. Premature scaling kills seventy-four percent of fast-growing startups that collapse. And the real killer, at 42% percent, is building something nobody needs, no product-market fit. We close on how to use AI to de-risk demand before you build, simulated interviews, market analysis from real user language, and vibe-coding an MVP to verify demand instead of inventing it.

This episode is about a failure pattern that's harder to spot than a bad idea — the False Positive. Baroo was a Boston pet care startup that launched into a perfect storm of lucky circumstances, mistook them for market validation, and scaled before the foundation existed. The idea was good, the founder was experienced, the early numbers were incredible. And that's exactly what killed it.

We flip the table and play the investor for a whole episode. A first VC call really comes down to three questions wearing different costumes: why this, why now, and why you. Everything else, market size, traction, the size of the round, just folds into those three. We walk through the first sixty seconds (can the VC repeat your idea back?), why "cheaper AI inference" is a weak "why now," the brutal founder-market-fit test (would I quit my job to come work for you?), and why AI lowering the build cost actually raises the bar on traction. Plus the rule of thumb for the ask at Series A.

This episode is a post-mortem of Triangulate, a Harvard-founded dating startup that tried to match people based on their digital behavior rather than self-reported profiles — a smart idea executed in entirely the wrong order. We trace their trajectory from initial thesis through two named pivots (Wings and DateBuzz) to an honorable shutdown in 2011, and diagnose the real cause: testing solutions before validating problems. The deeper lesson is about founder psychology — the same overconfidence that gets a company started is what makes rigorous hypothesis testing so hard once momentum builds.

What channels actually work for a solo founder with a zero-euro budget? We rank every free marketing channel as a live tier list - S down to F - arguing each placement instead of reading a checklist. Founder-led content and answering in communities take the top spot: fastest trust, zero spend, and it compounds. Traditional SEO drops to the bottom - Google's zero-click era killed it, and what's left is GEO, getting cited by the AI engines. Short-form video, warm-base newsletters, and influencer swaps land in the middle; random cold email is a hard F. Throughout, the real unlock is AI doing the grunt work so one person can run two or three channels at once - but we make the case that AI ghostwriting is a dead strategy, and "AI blindness" is now a real factor. We close on why agents have no money, and why you still have to talk to real people in human language.

Quincy had a real idea — tailored-fit women's clothing delivered online — and a real customer problem. So why did it fail? This episode is a full post-mortem: we go through the four failure layers (value proposition risk, manufacturing complexity, marketing cost, and the funding gap), explain why entering a red ocean without capital is a structural trap, and end with the question every founder avoids — why do you keep hiring people weaker than you need?

We're tackling the question founders ask most: how much equity to give a co-founder. The honest answer is there's no fixed number — equal splits have climbed to about 46% of two-person teams, but equal is only right when the contributions actually are, and what protects you isn't the percentage, it's vesting. We walk through the four-year vest with a one-year cliff, the CEO premium, part-time and moonlighting splits, every flavor of vesting from milestones to acceleration, and the classic horror story where a friend walks at month four still owning half the company. Then the German specifics — GmbH versus UG, why there's no restricted stock out of the box, and how Good Leaver / Bad Leaver clauses do the job a Delaware buyback would.

Jibo was a social robot built at MIT, backed by $73 million in venture capital, and designed by one of the world's leading robotics researchers. It still failed — and when the servers shut down, the robot said goodbye to its users on its own. In this episode we go deep on one case: what Jibo was, what broke, and what patterns from this story show up in founders today. Two bugs in particular — building in isolation from users, and treating feature-building as a sales cure — are the real cause of most startup failures. And underneath both bugs is something harder to fix than a product: learned helplessness.

Two startups and one structural problem. We open with the numbers on Germany's shrinking self-employed population — and why Kleinunternehmer traps, a 30-year AI content compliance agreement, and investors demanding multi-city proof before writing a pre-seed check all point to the same root cause. Pitchwise, a Stockholm AI fundraising stack for early-stage European founders, is building real infrastructure for a real gap — and accidentally pointing at a bigger missing piece: the "time for equity" advisor culture that the US has and Europe doesn't. Standard Template Labs raised $49M to automate corporate IT ticketing with AI agents, and the case is almost too obvious — 80% of first-level IT support is a decision tree that AI can run faster and cheaper than a human, and whoever locks in enterprise clients first owns an extremely sticky data moat.

Four fresh funding rounds under the microscope. Symbiotic Security raised $10M to automatically patch vulnerabilities in AI-generated code — their real moat is a security R&D team running an arms race with hackers, not the model itself. MINE raised $14M to bring conversational finance to Gen Z, but the unit economics of targeting low-ARPU users in a commoditized category are a tough climb. PraxisPro took $6M from AlleyCorp to train pharma sales reps with AI roleplay and call analysis — conservative use of AI in a domain where generic tools fail, and probably the strongest risk-adjusted bet in this batch. And Bolna raised $6.3M from General Catalyst to build India-specific voice agents for call centers, where the multilingual ASR problem is genuinely hard and the fast-iteration regulatory environment gives them an edge that European competitors don't have.