We started 1904 Labs because the way companies get built is changing — and the structures around them haven't caught up. AI compresses the time it takes to get from idea to working product, but the work of turning that product into a company still takes operators, not capital. So we built a studio for that work.
Three things we mean by AI-native startup studio:
- AI-native by default. Modern models live inside the products we ship — and inside how we ship them.
- Operator-built. We co-found, hands on the keyboard, from day one.
- Long-term, not handed off. We stay in the companies we start. There is no spinout date.
What we do, in practice
We co-found AI-native companies from zero. Sometimes founders walk in with a sentence. Sometimes the idea starts with us. Either way, we sit at the table from day one — strategy, product, engineering, GTM — and we stay long after the company can stand on its own.
We're three operators with three disciplines. Robert runs the studio's direction and sits with founders through the decisions that don't have a playbook. Garik is product lead and engineering lead in the same person — owning the technical core, the architecture, and the AI integration that has to actually run in production. Jaimie owns ICP, positioning, and the GTM motion.
What we look for in founders
We're looking for founders who will run the company we co-found together — long after the prototype, long after the launch. So we look for the things that compound across years:
- Domain experience that goes deeper than the deck.
- A version of the problem that has bothered them for a decade.
- Comfort with shipping before the answer is fully formed.
- An appetite for being told they're wrong about their own product.
Most founders we co-found alongside have lived inside the problem for years — accountants who got tired of how compliance gets done, ex-engineers who watched their team spend weeks on work that models could do in minutes, operators who've already exited once and want to do it again leaner. The thread is always the same: they understand the work, they want to change it, and they're willing to build the tools themselves.
How AI-native changes the work
When we say AI-native, we don't mean we sprinkled an LLM on top. We mean the products we build wouldn't exist — or wouldn't make sense — without modern models. The company's core unit of work is something a model does well, often better than the team doing it manually today.
That changes everything downstream:
- The product surface is smaller. Models do work that used to require a multi-screen workflow.
- The team is smaller. What used to take a ten-person engineering org gets done by three with the right tools.
- The economics are different. The marginal cost of doing the work scales with tokens, not headcount.
We don't think every company should be AI-native. But the ones that are get to skip parts of the early company-building playbook that no longer apply. That's the bet.
Why this works
Most early-stage companies don't fail from lack of resources. They fail from lack of focus, from technical decisions that don't survive year three, from products built without a GTM plan. The studio model — operators co-founding alongside founders, plural disciplines under one roof, in for the long run — exists to solve those failure modes.
We're based in San Diego. We're already shipping with founders. If you're building something AI-native, we'd love to hear from you.