Build things that matter.
Then throw everything you have at it.
How we see the world today
Everyone thinks it got easier. Cheaper to build, faster to ship, no gatekeepers left, so it must be easier to make money, right? Maybe at the margins, on speed to first dollar. But to build something that lasts? That wins a market? That defines a category and builds its own gravity? We think that’s harder now than it’s ever been. So fucking difficult. And as a founder, if you don’t feel that in your gut, you’re already headed down a path we can’t join you on.
Here’s why. When building costs and barriers go to zero, building stops being the thing that sets you apart. Everyone can do it. What used to take a team and a war chest now takes a weekend and a subscription. For you, and for every other person chasing the same thing you are. So the only two things left that matter are these.
Conviction, in a problem worth obsessing over.
Great founders carry a durable conviction about a problem. A belief that doesn’t move with every technology cycle or market trend. It usually predates the company. The company is just the vehicle. Bezos bet everything on the internet remaking commerce, and on radical customer obsession, before Amazon existed, and never wavered. Musk was convinced humanity needed sustainable energy and a path off the planet years before Tesla or SpaceX. The conviction came first. A read on where the world was going that the market hadn’t caught up to yet.
So ask yourself. What do you know that others don’t? What do you believe that gets laughed out of the room, too outlandish, too early, too unsexy to bother with? Bolting a 1% twist onto the obvious (“customer-support AI agents!”) wins you nothing that lasts. The companies that matter almost never look obvious at the start. Flexport bet that software and operations could remake freight forwarding, the least glamorous corner of global trade. Square bet the merchants banks wouldn’t touch, the food truck, the barber, were a market, not a rounding error. None of them were consensus. Each started from a belief the market didn’t share, and each ended up shifting, or creating, its category. That’s the pattern. A non-consensus bet to change something, run hard enough to become generational.
A big vision doesn’t mean a big product, a grand launch, or nailing PMF on day one. It means you see something bigger than where you start. You feel it, and you run at it. The first move is almost always a wedge. Small, sharp, deliberately narrow, because that’s how you get moving. Stripe’s wedge was seven lines of code to take a payment. The conviction underneath was enormous, and against consensus: that the internet was about to become the economy, that it was choking on financial plumbing built for a pre-internet world, and that developers, not banks, would rebuild it. The wedge was where they stood. The belief was where they were going. And if you’re still at the idea stage, that bigger thing is exactly what you’re hunting for. Not a feature, but something worth pointing your life at.
And it doesn’t have to arrive at us fully formed. Some teams are early, too early for that. But the drive toward it does. Twitter fell out of Odeo, a podcasting company, and the founders found the real thing by building. Conviction can form, deepen, and sharpen through real contact with the problem. What it can’t be is borrowed, safe, or missing. Stay adaptable to the solution. Hold the line on the bet, where this goes, and why you.
Which is why the deepest trap today is the tempting shortcut. Outsourcing that conviction to AI. Spar with it, pressure-test with it, let it challenge your assumptions, please do. But original thinking needs an original source. We’re all thinking with the same machine now, the same models, trained on the same averaged-out record of everything anyone has ever said. Ask it what to build, how to sell, what to believe, and it hands you the median. Source your worldview from there and, by definition, it’s everyone’s. And if it’s everyone’s, it’s worth nothing.
An LLM can sharpen your conviction. It cannot build it for you.
Distribution, speed of learning.
How fast can you move? Not motion for its own sake. Learning speed. How many real hypotheses can you test in a week? How fast do you update when the market talks back? How tightly do you close the loop between what you try and what you learn? That’s the compounding advantage. And in a market where someone can take you by storm overnight, you’re running for your life every moment of every day. Speed always mattered in startups. But outrunning a thousand others is a whole different thing than outrunning ten.
Conviction is the anchor. Speed is the engine, and the engine only matters when it’s pointed right. Founders close to the problem don’t just run more experiments, they run better ones. They know which hypotheses are even worth testing. That’s the compounding edge. Velocity aimed by judgment.
This is where AI earns its place. It compresses the cost and time of every experiment. More shots per week, faster synthesis, lower cost per insight. What it won’t do is tell you which experiments to run that aren’t yet in the data. That’s still yours. Feel, instinct, judgment.
Everything else, and we mean everything else, is secondary. By orders of magnitude.
What generational founders and companies are born now
Learn from history. Every technology cycle runs the same way. At the beginning, the ground is fresh and the generational companies get born, the ones that go on to define the era. The early internet gave us Amazon and Google. The early mobile cycle gave us Uber and Airbnb. Incrementalism is a late-cycle behavior. It’s what’s left when the platform is mature and all you can do is bolt one more feature onto what already exists.
We are at the very beginning of the AI cycle. Which means the incremental companies aren’t the ones being built right now. The amazing ones are. Build something incremental at the start of a cycle and you’ve misread the moment. You’re playing the end-game while the whole field is still wide open.
So the companies born now come from founders running at something the market can’t see yet. A problem that has to be solved, something newly possible that nobody’s claimed. And refusing to blink. You might start small, one step toward the vision. But you’re uncompromising about where it’s going.
That’s what we mean by bad ideas. The stuff that sounds wrong right up until it’s obviously right.
Why we exist
Here’s what most people get wrong. At seed, great founders don’t necessarily succeed. Great founders, and great companies, with real direction and speed of learning, can still die. But they rarely die on conviction or product. They die on go-to-market. On which experiments they run, in what order, and how well they run them. Moving from your first customers to a scalable way to acquire dozens, hundreds, or thousands more is the whole game between now and your next round. Get the bets wrong on what that engine is and it doesn’t slow you down. It ends the company before the product ever gets a shot.
That’s the one place we go all in, and not with a generic playbook. You read the customer and the market better than anyone in the world. Where the pain is, where they light up, what they’re actually telling you. And you distill it into an insight nobody handed you. What you’re usually not, at least not yet, is a world-class GTM expert who instantly knows the move that insight implies: which channel to open, which top-of-funnel test aligns to that delight moment, what to measure and in what order. That translation, from customer insight to GTM experiment, is a craft of its own. So that’s what we run towards. We put an army of the best GTM operators in the world, people reinventing go-to-market and winning at it right now, on your problem, and they turn your read of the customer into the sharpest GTM experiment, in the right sequence. Marv, our GTM co-pilot, keeps the machine moving. Living inside your GTM, compressing the loop between their read and your execution, so learning cycles that used to take weeks, months, or quarters happen in days. Your read of the customer stays the anchor. If it says an experiment is wrong, it’s wrong. We help drive the GTM move and the speed. You bring the conviction. Speed is your learning rate. And we make sure the cycles are the right ones, run in the right order, and don’t drift when it matters most.
We’re at the start of the biggest cycle of our lifetimes. The average is free, and worthless. So find the valuable thing you actually believe, the thing you HAVE to do, and run like hell at it with everything you’ve got.
— BADideas