His startup powers OpenAI's Voice Mode. Last month, they became a unicorn. | Russ d’Sa, Co-Founder of LiveKit

A tech giant was preparing to crush LiveKit when OpenAI dropped a bombshell: they had been secretly using LiveKit's infrastructure to power ChatGPT's voice mode all along. That revelation transformed what looked like an existential threat into a unicorn valuation for Russ d'Sa's real-time communications startup.

The path to that moment reveals how the best enterprise infrastructure companies get built , not through grand visions, but by solving immediate problems so well that customers can't live without you. Russ d'Sa stumbled into LiveKit during the pandemic after his previous company's acquisition to Medium left him with time and technical curiosity about WebRTC. "LiveKit started , it's the first thing I've ever done that sort of happened organically without intentionally trying to start a company," d'Sa explains. While his previous four startups had been forced attempts at entrepreneurship, LiveKit emerged from genuine need.

The pandemic created perfect conditions for real-time communication infrastructure. Within weeks of launch, LiveKit was powering 911 dispatch centers nationwide through a company called Prepared, which later sold to Axon. That early traction led to a $7 million seed round from Redpoint, but the real validation came from customers who kept expanding their usage. The company had found product-market fit by becoming invisible infrastructure that just worked.

D'Sa's approach to fundraising reveals why most founders fail at investor negotiations. Instead of accepting whatever terms investors offered, he and his cofounder deployed a calculated good cop, bad cop strategy. "Deezy, my cofounder, would basically talk to the investors and negotiate with them and then say, 'you know what? Russ is just not willing to do this,'" d'Sa recounts. When investors pushed back on terms, his cofounder would position d'Sa as the immovable obstacle. "I'll just be the bad guy here. You need that."

This tactic works because it preserves relationships while maintaining negotiating leverage. The "good cop" cofounder can empathize with investors while the "bad cop" takes responsibility for hardline positions. Most founders either accept bad terms to preserve relationships or burn bridges with aggressive tactics. LiveKit's approach split the difference strategically.

The acquisition dynamics that d'Sa describes expose why most startup exit strategies fail. Founders imagine they can manufacture interest through cold outreach or bankers, but the reality is starkly different. "It's not like you can email corp dev at Uber and be like, 'hey, you know, we're 10 engineers. Like, do you guys wanna buy us?'" he notes. "In all situations where a company is sold, not bought, it's almost purely relationship driven." The difference between getting acquired versus being ignored often comes down to having the right champion inside the acquiring company.

This relationship-driven reality explains why LiveKit's ultimate success wasn't about perfect positioning or market timing, but about building something customers genuinely depended on. When that tech giant threatened competitive pressure, OpenAI's revelation that they were already using LiveKit infrastructure transformed the conversation entirely. The startup wasn't just another vendor , they were critical infrastructure for one of the world's most valuable AI companies.

LiveKit's journey from accidental pandemic project to unicorn infrastructure company offers a blueprint for enterprise startups in the AI era. The companies that will matter aren't the ones building flashy demos or chasing trends, but the ones solving fundamental technical problems that other companies can't or won't tackle. They become indispensable by being invisible.

Watch for infrastructure companies that customers discover organically rather than through sales processes. When engineers choose your product without procurement involved, when usage expands without expansion sales, when competitors threaten but customers stay anyway , those are the signals that separate real infrastructure from venture-funded science projects. Russ d'Sa built LiveKit by accident, but the accident only happened because he was solving real problems for people who desperately needed solutions.