Wispr Flow is a voice dictation app for people who write by speaking into any text field, on Mac, Windows, iPhone, and Android. It turns natural speech into punctuated, formatted text in email, chat, documents, and code, and it learns names and vocabulary from corrections and a personal dictionary. The San Francisco company was founded in 2021 by Tanay Kothari and Sahaj Garg. This launch reports two months of production use for Canto 1, the first speech model from the Wispr Advanced Interfaces Lab, which has been running inside Flow since September. Canto 1 is a real-time model trained on spontaneous dictation, including laptop microphones, background noise, and very short utterances. On Wispr's evaluation of about 10 hours of English Flow dictations from more than 2,300 speakers, it recorded a 3.4% word error rate, the lowest among the models tested, including models from Google, OpenAI, AssemblyAI, and Deepgram. On a separate three-hour challenge set of noisy, low-volume, and short audio, it led the real-time models in that comparison and placed second overall behind Gemini 3.1 Pro, a much larger model unsuited to low-latency dictation. It was fast enough to replace an earlier setup that ran several recognizers on every dictation and waited for the slowest result, and it was post-trained to use personal context so learned names and terms show up more reliably later. Wispr reports that desktop users now edit 13% fewer dictations, that dictations taking more than a second to appear are down 54% on desktop and 37% on mobile, and that people dictate 7% more words a day. Daily volume had been flat for two months before the rollout, then rose as users moved from short messages toward longer documents and replies. A successor is already in training at more than ten times Canto 1's scale, aimed at noisy rooms, conversations with multiple speakers, and languages beyond English.