Uploaded September 2026 | Updated September 2026, 2 weeks ago
Tanay Kothari is co-founder and CEO of Wispr Flow, a voice AI company he started with co-founder Sahaj Garg in 2021, building toward a Jarvis-like assistant that lets people interact with computers through natural speech instead of keyboards and screens. His younger brother, Jaisal Kothari, is co-founder and CEO of Gallium and was Wispr's first software engineer, joining as a Stanford freshman to help build the company's earliest voice dictation technology.
Anne Dwane sits down with Tanay and Jaisal at a Village Global event to trace Tanay's path from building and shipping nearly 50 apps in secret as a teenager, including one that hit 2.5 million users before Google sent him a cease and desist, to founding Wispr as a hardware company and pivoting entirely to software after watching users ignore the physical device in favor of the app alone. Tanay breaks down Wispr's "zero edit rate" metric, a stricter and more user-centered standard than the industry's typical word error rate, walks through the viral Porsche-giveaway marketing stunt that pulled in tens of millions of views, and explains why he personally sat in on the company's first 500 onboarding calls. The conversation also covers how Wispr thinks about competing with frontier AI labs entering voice, why the company stacks multiple layers of defensibility instead of relying on one moat, and Tanay's approach to keeping teams small and communication overhead low as the company scales. Jaisal opens the conversation with the story of growing up alongside Tanay's secretive late-night coding sessions and how that early exposure shaped his own path into tech.
[00:00] Cold open
[00:29] Welcome to the event; introducing Jaisal
[00:53] Growing up with a 7-year age gap
[01:42] The app that hit 2.5 million users and got a cease and desist from Google
[02:12] What their parents thought about the secrecy
[03:00] The movie that inspired Tanay: Iron Man and the dream of building Jarvis
[03:47] Jaisal's own path into tech
[04:59] Joining Wispr as its first software engineer
[06:15] Tanay arrives; the origin story of Wispr
[06:39] Starting the company with co-founder Sahaj Garg in 2021
[07:59] Building a wearable hardware device for silent speech
[08:54] The pivot: realizing users only wanted the software
[09:06] "Sometimes the best product decisions come from humility, not conviction"
[10:20] The long-term vision: technology that fades into the background
[11:41] Why agentic AI assistants still don't work yet
[13:03] Building toward Jarvis, one step at a time
[13:08] The "zero edit rate" metric and why it beats word error rate
[14:27] The viral Porsche giveaway marketing stunt
[16:32] Personally running the first 500 onboarding calls
[18:33] Competing with OpenAI and other frontier labs on voice
[20:07] Why moats need to stack, not stand alone
[21:21] Growing the ML team and hiring Amazon and Meta's former voice AI leads
[21:22] Why Wispr avoids small edge models
[22:35] How Wispr trains on individual user data and privacy tradeoffs
[24:25] Scaling the team without losing efficiency
[25:56] Why bigger teams don't mean proportionally more output
[28:22] Predictions: multiple AI agents, personal compartmentalization, and the future of Apple's role
Tanay Kothari is co-founder and CEO of Wispr Flow, a voice AI company he started with co-founder Sahaj Garg in 2021, building toward a Jarvis-like assistant that lets people interact with computers through natural speech instead of keyboards and screens. His younger brother, Jaisal Kothari, is co-founder and CEO of Gallium and was Wispr's first software engineer, joining as a Stanford freshman to help build the company's earliest voice dictation technology.
Anne Dwane sits down with Tanay and Jaisal at a Village Global event to trace Tanay's path from building and shipping nearly 50 apps in secret as a teenager, including one that hit 2.5 million users before Google sent him a cease and desist, to founding Wispr as a hardware company and pivoting entirely to software after watching users ignore the physical device in favor of the app alone. Tanay breaks down Wispr's "zero edit rate" metric, a stricter and more user-centered standard than the industry's typical word error rate, walks through the viral Porsche-giveaway marketing stunt that pulled in tens of millions of views, and explains why he personally sat in on the company's first 500 onboarding calls. The conversation also covers how Wispr thinks about competing with frontier AI labs entering voice, why the company stacks multiple layers of defensibility instead of relying on one moat, and Tanay's approach to keeping teams small and communication overhead low as the company scales. Jaisal opens the conversation with the story of growing up alongside Tanay's secretive late-night coding sessions and how that early exposure shaped his own path into tech.
[00:00] Cold open
[00:29] Welcome to the event; introducing Jaisal
[00:53] Growing up with a 7-year age gap
[01:42] The app that hit 2.5 million users and got a cease and desist from Google
[02:12] What their parents thought about the secrecy
[03:00] The movie that inspired Tanay: Iron Man and the dream of building Jarvis
[03:47] Jaisal's own path into tech
[04:59] Joining Wispr as its first software engineer
[06:15] Tanay arrives; the origin story of Wispr
[06:39] Starting the company with co-founder Sahaj Garg in 2021
[07:59] Building a wearable hardware device for silent speech
[08:54] The pivot: realizing users only wanted the software
[09:06] "Sometimes the best product decisions come from humility, not conviction"
[10:20] The long-term vision: technology that fades into the background
[11:41] Why agentic AI assistants still don't work yet
[13:03] Building toward Jarvis, one step at a time
[13:08] The "zero edit rate" metric and why it beats word error rate
[14:27] The viral Porsche giveaway marketing stunt
[16:32] Personally running the first 500 onboarding calls
[18:33] Competing with OpenAI and other frontier labs on voice
[20:07] Why moats need to stack, not stand alone
[21:21] Growing the ML team and hiring Amazon and Meta's former voice AI leads
[21:22] Why Wispr avoids small edge models
[22:35] How Wispr trains on individual user data and privacy tradeoffs
[24:25] Scaling the team without losing efficiency
[25:56] Why bigger teams don't mean proportionally more output
[28:22] Predictions: multiple AI agents, personal compartmentalization, and the future of Apple's role
