Voice AI startup Wispr has raised a $25 million round from Notable Capital while its dictation app Flow is ramping up with consumers and enterprise users. The investment adds veteran investor Hans Tung as an observer to the board and will be used for hiring, deeper speech modeling, and a swifter expansion of its platform, the company said.
Traction points for changing workplace workflows
Overall, on average people are now writing more than 50% of their characters through Flow after using it for three or more months, Wispr reports. Since June, usage has increased some 40% month over month, helping propel a roughly 100x increase in user base annually with 70% retention over 12 months.
On the enterprise side, the company says it has penetrated 270 of the Fortune 500 list and signed up 125 companies as its customers. That footprint hints that voice input may be transitioning from a fun, niche productivity hack to an everyday part of the workflow for knowledge workers.
Wispr did hit a learning curve as less tech-savvy folks started showing up. A lot of people had tried dictating within the app, not realized that it worked elsewhere on their phones, and churned. A new onboarding flow now helps users get up and running with dictation across their favorite apps, and that small UX change had a real impact on activation by funneling more people to launch dictation, the company says.
Gartner and IDC have both pointed to voice interfaces as an increasingly key input modality within productivity suites. Wispr’s measure of “share of characters” is a neat way to quantify that behavior on the part of users, shifting from downloads and into long-term habit formation.
Why Notable Capital is leaning in on Wispr’s growth
Wispr’s leadership said Notable Capital had invested in the company on deep diligence that included competitor interviews and product teardowns. Credit: Wispr. Tung—whose past investments include Affirm, Airbnb, Slack, Coinbase, Anthropic, and TikTok—has arrived as a board observer following an investment that suggests belief in Wispr’s potential as a multiplatform interface product.
The wager also mirrors a wider venture trend toward distribution-first AI. Just as Grammarly turned itself into a default writing layer by building a home inside every text field, Wispr wants to be the ambient voice layer for any app where text is produced.
Wispr’s technical edge today and the product roadmap
Accuracy is the battleground. Wispr purports to have an average error rate of around 10%, versus 27% for OpenAI’s Whisper and 47% for Apple’s own transcription in its internal tests. Word error rate—advocated by the National Institute of Standards and Technology, for instance—is still the standard, but time-to-correct and personalization can often be more important at work: The fewer edits, the faster throughput.
To get there, Wispr envisions customized automatic speech recognition that’s trained on each user’s voice, domain jargon, and writing style. Flow currently supports Windows, Mac, and iOS; an Android beta is planned by the end of this year, with a stable release expected in Q1 2025.
Yet there are a few “actions,” or Siri Shortcuts, you may have set up that made Flow-roulette more unpredictable—called instead things like Draft Email and Reply to Email.
(Image source: Runner’s High)
All of which is to say: the roadmap has always aimed beyond dictation for ways of incorporating speech input with other actions, 1 circling from single play toward co-piloting your AI assistant. Wispr is also testing a private API with key enterprise and hardware partners, and anticipates wider access for developers next year.
Competitive landscape for voice AI and dictation tools
The category is crowding fast. YC-backed Willow and Aqua, Monologue from the Every ecosystem, and tools like Typeless, Talktastic, Superwhisper, and Betterdictation all want mindshare. Meanwhile, the platform incumbents—Apple, Google, Microsoft—work on leveraging native voice features, and model providers like OpenAI and Anthropic strive toward E2E speech pipelines.
In this world, winners will compete on accuracy across accents and specialized vocabularies, latency, privacy controls, and perhaps above all on true cross-app distribution. HCI researchers at universities like Stanford and MIT have long espoused that the best interface is the one users hardly perceive—the product ethos of Wispr is squarely targeted there.
What the new funding enables for Wispr and Flow
The new capital will be used to hire top machine learning and speech talent who might otherwise be lured by OpenAI or Anthropic, as well as to support national growth and build hardened enterprise features. If Flow can capture more than 50% of the characters of a user, it becomes the natural input habit—a sticky moat with switching costs that aren’t contractual but behavioral.
Key milestones to look for:
- The Android rollout
- An open API
- Enterprise case studies that document productivity gains
- Progress around personalization that cuts post-dictation editing
If Wispr can keep adoption rates high and error rates and latency low, it has a real chance to own the voice-to-text layer across the modern workplace.