Kulbir Singh, Founder and CEO of Rexpt, joins Seth Narayanan to discuss why voice AI is poised to become the next dominant computing interface. With roots in machine learning dating back to 2014 and an early startup exit to Dell, Kulbir brings a practitioner’s perspective on how AI has evolved from niche automation to real-time, decision-making agents that can run critical business workflows.
The conversation explores the creation of Rexpt, an AI receptionist platform that can deploy in minutes, as well as the broader implications of agentic AI for organizations. Kulbir shares how voice systems are moving beyond simple triage into full transactional capabilities — capturing leads, processing payments, and managing customer interactions autonomously.
Listeners will learn why experimentation beats experience in AI hiring, how teams are using multiple AI models collaboratively, and why humans will remain essential even in a highly automated future.
“I belonged to a small town… computer was pretty new… but I took every opportunity that came my way.”
Kulbir’s journey began with a first encounter with computers in rural India, eventually leading to founding a company and building machine learning systems years before the AI boom.
His early email-analysis engine — later acquired by Dell — revealed how startups could create massive value through focused innovation.
“Voice AI… is the new interface… the next thing in how we interact with mobile apps.”
Kulbir argues that voice is not just another feature — it’s the future of human-computer interaction. Modern systems can now handle real-time conversations, complete transactions, and adapt to business workflows without requiring screens or forms.
“It will simply ask you all the details… send you a payment request… and set up everything for you on the phone.”
Early voice bots triaged calls. Today’s agents execute tasks end-to-end — qualifying leads, filling forms, and sending payment links. This shift marks the transition from automation to autonomy.
“What used to take 20–25 days… we are doing in less than four hours.”
Kulbir shares how AI agents are already transforming internal workflows — compressing documentation processes from weeks to hours. Multiple models collaborate, with humans validating outputs — a hybrid approach that balances speed with accountability.
“AI is a tool… humans will always be required to do certain things.”
Despite rapid progress, Kulbir rejects the idea of a fully automated world. The future worker, he argues, will operate more like a team lead — orchestrating AI collaborators rather than performing routine tasks.