Wispr just raised $280 million selling something your phone already does for free

Every phone sold in the last decade can turn your speech into text for free. A company that does the same thing was just valued at $2 billion. The gap between those two facts is the whole business.

On August 17, 2026, Wispr said it had raised $280 million in a Series B led by Menlo Ventures at a $2 billion valuation, bringing total funding to $361 million. Returning backers included Notable Capital, NEA and 8VC, with new money from Acrew, Forerunner, Goodwater and Peak XV. The round landed less than ten months after a raise that valued the company at $700 million, roughly tripling that price.

The product is called Flow. You speak into any text field on your computer or phone, and text appears. Which raises the obvious question: Apple, Google and Microsoft all ship dictation for free, and have for years.

The insight most coverage of this round skips is that Wispr is not really selling speech recognition. It is selling the removal of the edit. The company's internal success metric is not accuracy in the usual sense. It is what Wispr calls the zero edit rate, the share of output you do not have to touch at all. Free built-in dictation transcribes what you said, which includes your "ums," your false starts, the sentence you began badly and restarted, and no punctuation you did not speak aloud. What you get back is a transcript you then have to clean up, which is why almost nobody uses it for real work. Flow's pitch is that it deletes that step: it strips filler, fixes stumbles, and formats the result into prose you can send. Kothari's framing is that the point of speaking instead of typing is "to stay inside your own train of thought," and a cleanup step breaks exactly that.

The commercial evidence is unusually stark.

Above are the numbers of a company that solved one specific irritation well enough that people pay rather than tolerate it. More than 60 billion words have now been written through Flow, across 10,000-plus enterprises.

Where that insight came from is the better story.

Wispr was founded in 2021 by two Stanford roommates, Tanay Kothari and Sahaj Garg, to build something far more ambitious: a non-invasive neural wristband that would read nerve signals so you could mouth words silently and have them appear as text. They spent more than two years on it. Kothari's own account of that period is blunt: "we've made every mistake imaginable, including building a brain-computer interface our own team didn't want to use." In 2023 they looked at the software layer built to make the wristband's output usable, the part that turned messy input into clean text, and realized that layer was the actual product. They shipped it as a Mac app in October 2024, then Windows, iOS and Android.

"We've made every mistake imaginable. Including building a brain-computer interface our own team didn't want to use. But at year 3, we pivoted and launched Wispr Flow." - Tanay Kothari, co-founder and CEO, Wispr

And now the money is going back to the moonshot.

Alongside the round, Wispr announced Canto, a speech model trained on messy real-world audio rather than clean recordings, which it says cuts word error rates in noisy settings from over 30% to 5 to 10%. It launched a meeting notetaker last week, putting it against Granola, Fireflies and Read AI. And it opened the Wispr Advanced Interfaces Lab under chief scientist Ariya Rastrow, previously on Amazon's founding Alexa team, alongside a hardware partnership with the Oasis ring for quiet dictation. The dictation app is funding a return to the interface research the company abandoned.

The honest catch. Wispr does not publish revenue. The last solid public figure is roughly $10 million ARR in October 2025, with Kothari claiming 150x revenue growth in the year to May 2026, a claim no audited number backs. The customer experience is also more contested than the funding headline suggests: Wispr's Trustpilot rating sits around 2.7 out of 5, with a recurring complaint that accuracy degrades once the free trial converts to a paid plan, plus concerns about AI cleanup occasionally rewriting what someone actually said. Canto reads, in part, as an answer to those complaints. And the category is crowded: Willow, Monologue, Aqua and Superwhisper chase the same wedge, before you count the free option in every operating system.

Wispr spent two years building the future of input and accidentally built a business fixing the present of output. The $280 million is a bet that it can now afford to go back and try the hard thing again.