Wispr (Wispr Flow)
Last reviewed
Jun 8, 2026
Sources
9 citations
Review status
Source-backed
Revision
v1 · 1,620 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
Jun 8, 2026
Sources
9 citations
Review status
Source-backed
Revision
v1 · 1,620 words
Add missing citations, update stale details, or suggest a clearer explanation.
Wispr (legally Wispr AI, Inc.) is a San Francisco artificial intelligence company best known for Wispr Flow, an AI voice dictation application that turns natural speech into clean, formatted text inside any application across macOS, Windows, iPhone, and Android. The product combines speech recognition with large language models to remove filler words, fix grammar, adapt tone, and learn a user's vocabulary, positioning voice as a primary input method that the company frames as a "Voice OS." [1][2] Founded in 2021, Wispr originally set out to build a non-invasive neural wearable for silent text entry before pivoting in 2024 to the Flow software application. [3] By 2026 the company reported roughly 2.5 million downloads and use by employees at around 270 Fortune 500 companies, and in May 2026 it was reported to be in talks to raise approximately 260 million dollars at a valuation close to 2 billion dollars, led by Menlo Ventures. [4][5]
Wispr Flow is an always-available dictation layer that runs system-wide rather than inside a single app. A user holds a hotkey (or, on mobile, uses a dedicated keyboard), speaks naturally, and Flow inserts polished text wherever the cursor sits, whether that is an email client, a messaging app, a code editor, or a chatbot. [2][3] Unlike raw transcription, Flow uses language models to clean up speech in real time: it strips "um" and "uh," repairs false starts, applies punctuation and formatting, and can adapt register from a casual Slack message to a formal email. [1][2]
The company markets Flow as roughly four times faster than typing, citing about 220 words per minute by voice versus about 45 words per minute on a keyboard, and supports dictation in more than 100 languages (reported as 104) with automatic language detection. [1][2] Additional features include AI Auto Edits, a personal dictionary that learns names and jargon, a snippet library of voice-triggered shortcuts, and a command mode that interprets spoken instructions as natural-language commands rather than literal text. [1] Settings and dictionaries sync across platforms. Wispr has stated that Flow's word error rate is around 10 percent, which it contrasts with higher rates it attributes to general-purpose transcription models and native operating-system dictation. [6]
Adoption skews broader than developers: the company reports that over 30 percent of users come from non-technical backgrounds, and it highlights uptake among people with ADHD, dyslexia, carpal tunnel syndrome, and paralysis, for whom typing is slow or painful. [3] Roughly 40 percent of dictations are in English, with the remainder spread across languages such as Spanish, French, German, Dutch, Hindi, and Mandarin. [3]
Wispr was founded in 2021 by Tanay Kothari (chief executive officer) and Sahaj Garg, who met as students at Stanford University. [3] Kothari had previously worked at Stanford's AIMI medical-AI lab, co-authoring research on detecting pulmonary embolism in CT scans, and as a teenager had built consumer products including a music-intelligence service. [7]
The company's original ambition was hardware: a non-invasive wearable that would let people control a smartphone without touch by mouthing words silently, with the device translating the associated neural and muscle signals into text. [3] After roughly three years of development, the founders concluded that the contemporary state of AI, hardware miniaturization, regulatory clearance, and the level of user trust required made a consumer brain or silent-speech interface impractical at scale. [3][8]
The pivot, in 2024, came from a realization that the most valuable piece they had built was not the hardware but the software layer originally created to support it: a dictation engine that cleaned up speech, removed filler, and formatted text automatically. That layer became Wispr Flow. [8] The Mac application launched in October 2024, followed by Windows in March 2025 and an iOS app in June 2025 that functions as a third-party keyboard so voice can be used in any mobile app; an Android version followed. [2][3] In the roughly 18 months after the pivot, the company reported growth of about 40 percent month over month in both users and revenue. [9][6]
The table below summarizes Wispr Flow as described by the company and contemporaneous reporting.
| Attribute | Detail |
|---|---|
| Product type | System-wide AI voice dictation that types into any app |
| Platforms | macOS (Oct 2024), Windows (Mar 2025), iPhone (Jun 2025), Android |
| Languages | 100+ (reported 104), with automatic detection |
| Claimed speed | ~220 words per minute by voice vs. ~45 typing (~4x) |
| Stated word error rate | ~10 percent |
| Key features | AI Auto Edits, command mode, personal dictionary, snippet library, cross-device sync |
| Pricing | Free tier; Flow Pro subscription (14-day free trial) |
| Vision | "Voice OS": voice as a primary computing input and assistant |
Technically, Flow pairs automatic speech recognition with language models that adapt to individual users over time, learning vocabulary and preferred style. [3] The company has said it is building proprietary, personalized speech-recognition models rather than relying solely on third-party transcription, and that it began testing a closed enterprise API with select partners in 2025 ahead of broader developer access. [6] Wispr's stated roadmap extends Flow beyond dictation toward a "voice-led operating system" that can initiate workflow automation such as drafting and sending email, with longer-term plans to evolve Flow into a broader AI assistant for messaging, note-taking, and reminders. [3][6]
Wispr's funding history reflects its shift from hardware to a fast-growing software product. The company raised early backing from investors including Neo, NEA, 8VC, and AIX Ventures during its hardware phase. [3]
After Flow launched, Wispr raised a 30 million dollar round led by Menlo Ventures in June 2025, with participation from NEA, 8VC, and angel investors including Pinterest co-founder Evan Sharp and Carta chief executive Henry Ward; reporting at the time described total funding of about 56 million dollars. [2] In November 2025 the company added a 25 million dollar extension led by Notable Capital, with participation from Steven Bartlett's Flight Fund, at a post-money valuation of 700 million dollars, bringing total funding to about 81 million dollars. [6]
In May 2026, Bloomberg reported that Wispr was in talks to raise approximately 260 million dollars at a valuation close to 2 billion dollars, in a round expected to be led by Menlo Ventures. As reported, the deal was not finalized and terms could change; the proposed valuation would represent close to a tripling of the 700 million dollar figure set about six months earlier. [4][5] The reported and in-progress rounds are summarized below.
| Date | Round | Amount (USD) | Lead | Valuation |
|---|---|---|---|---|
| Jun 2025 | Series A | ~30 million | Menlo Ventures | Not disclosed |
| Nov 2025 | Series A extension | ~25 million | Notable Capital | ~700 million (post) |
| May 2026 (reported, in talks) | New round | ~260 million | Menlo Ventures | ~2 billion |
Reported traction underpinning the 2026 round included around 2.5 million downloads, use by employees at roughly 270 Fortune 500 companies (with named users at firms such as Nvidia and Amazon), about 70 percent retention over 12 months, and roughly 3.8 million dollars in revenue between July 2024 and July 2025; the company also said it was signing on the order of 125 new enterprise customers per week in late 2025. [3][4][5][6] These growth and revenue figures come from the company and from press reports rather than audited disclosures and should be read as such.
Wispr competes in a crowded and fast-moving category of AI dictation and voice-input tools. Direct rivals named in coverage and by the company include Aqua Voice, Superwhisper, and Talktastic, alongside meeting-focused transcription products such as Otter.ai and the built-in dictation features of macOS, Windows, and iOS. [3][2] More broadly, Wispr sits within the wider voice user interface and voice-AI landscape that includes speech-synthesis and voice-agent companies such as ElevenLabs, though those address generation and agents rather than personal dictation.
Wispr's significance lies in its bet that voice can become a mainstream primary input method for knowledge work, not just an accessibility feature or a meeting-notes tool. By layering large language models on top of speech recognition to produce edited, context-aware text, Flow exemplifies a shift from literal transcription toward intent-level dictation, an approach echoed across the 2024 to 2026 wave of AI writing and productivity tools. Whether the company can sustain its reported growth, defend against native operating-system dictation that is improving with on-device models, and justify a roughly 2 billion dollar valuation remains the central open question, but its rapid enterprise adoption and the hardware-to-software pivot have made it a closely watched example of voice-first computing. [4][9][8]