Voice AI startup Wispr Flow closed a $280 million Series B at a $2 billion valuation on Monday, betting that its new speech model Canto can make dictation reliable enough for professional work.
The round and the rationale
The financing, announced August 17, arrives as the company shifts from consumer voice assistants, weather queries, timers, to high-stakes dictation where errors break a user's train of thought. Chief executive Tanay Kothari framed accuracy as the sole gatekeeper for adoption, saying the largest portion of the proceeds will fund that effort. The round's structure, including consideration type and any break fees, was not disclosed.
The model and the claims
Wispr Flow is previewing Canto, a proprietary speech model built for noisy environments such as cars, open offices and public spaces. The company says error rates in the hardest conditions drop from above 30 percent of words to between 5 and 10 percent, and that everyday dictations requiring edits should fall by 30 to 35 percent. Chief scientist Ariya Rastrow, a founding member of the Alexa team, leads the research effort.
The B2B pitch and what follows
PYMNTS reported last month that voice AI is moving from comprehension to execution, collapsing multi-step workflows into single conversations for warehouse supervisors, field technicians and procurement managers. Wispr Flow is positioning Canto as the accuracy layer underneath that compression. The next test is whether enterprise buyers trust the benchmarks enough to embed voice in core processes, and whether the $2 billion price tag holds if adoption curves flatten.