EN
Back to the archive

The archive · Consumer Apps · Product decision · 2021–2026

Wispr bets voice replaces the keyboard; $280M Series B at $2B after 30x revenue growth

A Stanford pair spent 3 years on silent-speech tech, pivoted to an AI dictation layer in every app, and raised $280M at a $2B valuation.

Wispr Flow

The betThat voice, not the keyboard, becomes the primary interface between people and AI — starting with dictation in the humble text box, then actions, then wearables.Scaling

What the business is

AI dictation layer that turns speech into polished text in every app on Mac, Windows and iPhone, plus an AI meeting notetaker.

Starting capitalSelf-funded from a Stanford dorm; total funding reached $361M after the August 2026 Series B led by Menlo Ventures.

How it started

Tanay Kothari and Sahaj Garg met in a Stanford dorm; inspired by Iron Man's JARVIS, they founded Wispr in 2021. They spent about three years on silent speech, wearables and brain-computer interfaces — demos that wowed but never became daily habits. So they retreated to the input box and made dictation the wedge.

What happened

Flow's 2024 Product Hunt launch hit #1 with 2.1k upvotes, and adoption spread bottom-up into 125,000+ businesses and most of the Fortune 500. In August 2026 Wispr raised a $280M Series B at a $2B valuation led by Menlo Ventures (total $361M), reported ~30x revenue growth in a year, and released Canto, its first in-house speech model, to answer accuracy complaints.

How it ended up

Still scaling: expanding from dictation into Notetaker and the Advanced Interfaces Lab, with a roadmap from voice input to voice-triggered actions to wearables — racing to own the layer before OS giants make voice a default feature.

Background

Wispr Flow is an AI dictation layer that turns natural speech into polished text inside email, docs, chat, the terminal and AI tools — no separate app to open. The company's real bet is bigger than dictation: that voice replaces the keyboard as the primary way people hand intent to software, and that the startup that owns that layer wins the post-keyboard era.

Co-founders Tanay Kothari and Sahaj Garg met in a Stanford dorm and founded Wispr in 2021. Inspired by JARVIS, they first spent about three years on silent speech, wearables and brain-computer interfaces — impressive demos that never became daily habits. Their pivot looked like a retreat: stop building the assistant, start with the humble text box, where typing happens hundreds of times a day.

The wedge worked. Flow's 2024 Product Hunt launch hit #1 Product of the Day with 2.1k upvotes, and employees pulled it into 125,000+ businesses, most of the Fortune 500. In August 2026 Menlo Ventures led a $280M Series B at a $2B valuation, taking total funding to $361M; Wispr released Canto, its own speech model, and framed the next steps as voice-triggered actions and wearables.

What has to be true

  • The pivot from silent speech to dictation kept the long-term bet (voice replaces keyboard) while fixing the short-term usage problem.
  • The input box is the highest-frequency human-computer interaction, turning Flow from a demo into a daily habit.
  • Bottom-up adoption across 125,000+ businesses and most of the Fortune 500 created distribution before any enterprise sales team existed.
  • Model-neutrality — working across every app and every AI tool — positions Wispr as the interface layer rather than a feature of one OS.

What can be applied

A futuristic vision needs a boring wedge: Wispr spent three years on silent speech, then won by putting voice in the one box people use hundreds of times a day — the input field.

Aftermath

As of September 2026 Wispr is scaling on the $361M it has raised: Flow runs on Mac, Windows, iPhone and Android, Notetaker is bundled in, the Canto speech model is deployed, and the Advanced Interfaces Lab (led by former Alexa founding member Ariya Rastrow) is pursuing context-aware voice actions. Its stated risk is the window before Apple, Google and OpenAI make voice a default OS feature — at which point Wispr must prove the gap between a professional layer and a free feature.

Sources

spotted an error? The archive wants to know.

Your turn

You just read one. Describe what you are building, and see who is betting on the same thing.

Free account · 3 free questions · no card

Related cases