Harshil Mathur started Razorpay after quitting the highest-paying job on his campus, a role his whole family had just celebrated, because he walked in on day one and realised he was a guy who wanted to sit and code, not step onto an oil field.
Then he spent a decade away from that: walking into bank after bank getting laughed out of the room, surviving the grind no funding can fast-track, and the night Yes Bank froze with customer money stuck inside it. This is the founder story, lived experience as an edge, why the rejections compounded in his favour, why the grind always comes, and the values that made the hard calls simple.
And then the thing that pulled him back: agentic AI. "It went from being an assistant to an execution engine." Six years after he last wrote real code, Harshil locked himself in a room, asked "if I were to start Razorpay today, how would I build it?" — and rebuilt everything.
The second half is an operator's view of what that shift actually changes:
1. Why AI magnifies an org's weaknesses instead of fixing them
2. Why an agent with no plan drifts exactly like a company with no plan
3. How Razorpay flipped its leadership hackathon and the bet behind Agent Studio
4. Hosted by Avnish Bajaj with Vikram Vaidyanathan this is a conversation about building, walking away from it, and being pulled back, and what that says about where AI is headed.
%20(1).avif)








.webp)
.avif)
%20(1).avif)


