Ramp keeps startup speed at $44bn by tolerating failure
Ramp tries to keep startup speed at a $44bn valuation by treating failure as a cost of scale. It funds small bets, runs independent pods and hires people with agency, because coordination costs grow with size. Bonaci applies the same logic to AI, which he expects to create more work and more hiring rather than less.
Takeaways
- Bigger companies must tolerate more failure. Coordination costs grow with size, so Bonaci says leaders must let individuals move independently. Ramp's fortune cookie ad test in restaurants, which he first thought would fail and could cost hundreds of thousands of dollars, drove measurable traffic to ramp.com.
- Small single-threaded pods keep Ramp a startup. Ramp prizes velocity in every function and accepts duplicated roles. A product manager proves product-market fit for new products before growth joins. The accounts payable pod began as one person and became a project lasting over a year.
- AI adopters grow faster and hire more. Ramp research shows companies adopting AI grow faster and hire more, and top adopters hire more junior people. Bonaci cites Jevons paradox, where cheaper work means more work, and argues against automating tasks that train juniors.
- AI users show no model loyalty. Users default to the most intelligent, expensive model and waste money, so Bonaci argues for routing tools. Both speakers feel overloaded by running many conversations and agents, and he compares now to the early internet and dot-com boom.
- Markets misjudge AI disruption of software. Bonaci bought Figma stock after the market overreacted to the launch of Claude Design, and says markets forecast disruption badly. Ramp mostly uses AI to build its own tools rather than replace paid software, prioritizing growth over savings.
- Hiring and culture are the bet Ramp cannot miss. Bonaci favors one-page documents and builders who show their work in interviews. Reluctance to join a project usually signals a hiring mistake. AI will commoditize routine tasks, while high-trust, creative and expert work stays with people.
Bonaci's playbook for funding bets
- Set aside a budget with no return expectations so teams can back unproven ideas.
- Scope each experiment to the smallest budget that gives a signal.
- Prefer fast learning to slow perfection.
- Judge most bets by the conviction of the person pitching them.
- Expect senior leaders to be right more often than not, while early-career staff get leeway to fail more than 90 percent of the time if they are learning.
Where Bonaci sees buying heading