Last month, in the corner of a tech meetup, I seized an opportunity to chat with Kite’s founder, Chi Zhang, for quite a while. To be honest, the environment was pretty noisy—people everywhere discussing projects, the coffee machine whirring loudly. I asked him directly, “Everyone knows building an AI economic layer is a tough nut to crack, so why are you guys so determined to tackle it?”

His answer was pretty unexpected. He didn’t talk about changing the world or anything grand like that; instead, he brought up his previous experience working at Google Cloud and Snapchat. He said there’s an image he can’t forget—those world-class AI models, incredibly powerful, yet stuck waiting on servers for humans to give them commands. It’s like a bird that can fly, but is locked in a cage and never gets to take off.

That really made me understand. What brings together this group of Stanford and MIT graduates, along with core tech talent from Silicon Valley giants, isn’t chasing a hot trend for quick money—it’s about genuinely wanting to solve a fundamental problem: How can AI not only do tasks, but also make its own decisions and create value in the market independently?

From what I’ve observed, what’s most impressive about the Kite team is their ability to “mix and match.” They’re not just pure smart contract developers from the crypto world, nor are they academics locked in a lab writing papers—they’re more like “battle-tested architects” forged on the front lines of business.

Their CTO, Vibhav Bhargava, used to be responsible for large-scale systems at Meta, and that experience is absolutely critical. The whole team’s approach to design is grounded—they don’t start by theorizing about maximum throughput, but instead ask questions like, “During a big shopping event, how fast does an AI shopping assistant need to react to help a user snag a flash deal?” This way of working backward from real-world scenarios to define technical requirements is a totally different mindset from projects that just pile up numbers in their whitepapers.

This also explains why their testnet could handle 17 billion interactions and remain stable. It’s not luck—it’s that from day one, they designed with real user behavior and edge cases in mind.

Simply put, building economic infrastructure for AI requires more than just technical idealism; you need experience dealing with challenges in real business environments. The most valuable thing about the Kite team may just be this hybrid skill set—they understand both AI and business, can write code, and know user pain points.