Wispr raises $280 million at $2 billion valuation for voice AI

Voice AI startup Wispr raised $280 million at a $2 billion valuation, nearly tripling its prior mark, as investors bet voice becomes the primary interface for AI.

Wispr raised $280 million at a $2 billion valuation, nearly tripling its prior mark, as investors bet voice becomes the primary interface for AI. The Series B, led by Menlo Ventures, brings total funding to $361 million and arrives six months after an oversubscribed Series A2.

"The bottleneck in AI has moved from the model to the human interface to the model, and Wispr is the interface," Matt Kraning, partner at Menlo Ventures, said. "We watched Flow spread through most of the Fortune 500 before there was a sales team to speak of, pulled in by employees one desk at a time."

Flow, Wispr's flagship dictation product, is now used by millions of people across more than 125,000 businesses, including employees at most Fortune 500 companies. Revenue grew more than 150 percent in each of the past four quarters. Alongside the raise, Wispr previewed Canto, its first proprietary speech model, which cuts word error rates from more than 30 percent to roughly 5-10 percent in noisy conditions, wind and strong accents.

The capital funds the Wispr Advanced Interfaces Lab, led by Ariya Rastrow, a founding member of Amazon's Alexa team and former multimodal foundation model lead at Meta. Wispr plans to extend voice beyond dictation into wearables and other hardware, competing with meeting-note tools Granola, Fireflies and Read AI.

Canto targets the accuracy gap in real-world speech

Traditional speech recognition systems are trained on controlled benchmark audio, which fails in the noisy environments where people actually dictate. Canto is trained on real-world usage data and contextual signals from Flow, allowing it to handle background noise, wind and strong accents. Wispr said users would need to edit roughly 30-35 percent fewer dictations as a result. The company did not disclose the test conditions for the comparison.

The model marks a shift for Wispr from an application-layer company to one building its own underlying AI technology. Owning the speech model lets Wispr optimize it around the specific environments and interaction patterns generated by Flow, rather than relying on third-party transcription engines. The launch follows weeks of user complaints about a quality dip in Flow's dictation output, which the company has attributed to the transition to Canto.

Voice moves beyond the text box

Wispr's thesis is that AI model intelligence is advancing faster than the interfaces people use to communicate with those models. While AI systems can understand complex instructions, most interactions still begin with typing into a text box. Wispr sees voice as a way to reduce the friction between thought and expression.

The company's research lab is investigating wearables, where keyboards and screens are less practical. Wispr has also partnered with hardware makers like the Oasis ring to let customers dictate without speaking loudly, and launched a meeting notetaker that competes with Granola, Fireflies and Read AI.

The dictation market is getting crowded. Apps like Willow, Monologue, Aqua and Superwhisper target prosumers at lower price points, while Nuance, owned by Microsoft, dominates healthcare dictation and Otter.ai has carved out meeting transcription. Deepgram offers developer-friendly speech APIs. Wispr's differentiation will need to be profound to justify its valuation premium, requiring either breakthrough accuracy or a platform play that turns voice into infrastructure rather than a feature.

Wispr's valuation of $2 billion puts it in rare company for a voice AI specialist, roughly double what many enterprise SaaS companies command at similar stages. The company's enterprise adoption has been organic — pulled in by employees rather than a traditional top-down sales team — which Menlo Ventures cited as evidence of product-market fit. The round also attracted athletes and cultural figures including Livvy Dunne, Shaun White, Dak Prescott and Klay Thompson, a sign of the consumer appeal Wispr is building. Whether Wispr can convert its dictation traction into a broader platform for voice-driven human-computer interaction will determine if the valuation holds.

This article is for informational purposes only and does not constitute investment advice.