Google is adding Gemini-powered dictation to Gboard, the keyboard app that ships on hundreds of millions of Android devices. The feature launches first on Samsung Galaxy and Google Pixel phones, with broader rollout likely to follow.
For users, this means AI-quality voice-to-text without paying for a separate app. Gemini handles the transcription, which should produce more accurate results on complex sentences, technical vocabulary, and varied accents than older rule-based speech recognition.
The harder story is what this does to standalone dictation apps. Many tools in this category - including professional-grade voice apps that charge $8-15/month - have built their products around essentially this capability: AI voice to text that works inside any app. When that ships inside the keyboard you already have, the subscription argument becomes difficult to sustain.
Google has run this playbook before: identify a category where Android users pay for third-party apps, build the feature into the OS, and let developers figure out differentiation. It happened with call recording, navigation, and PDF reading. AI dictation is next on that list.
The tools most exposed are those competing primarily on accuracy and convenience. The tools with a better survival story are those competing on depth: domain-specific vocabulary for legal or medical transcription, desktop-first integrations, enterprise security requirements, or team features Gboard will never build. Wispr Flow, which focuses on professional voice workflows across desktop environments, has a clearer differentiation story than a basic Android app trying to out-accurate Google's keyboard.
Developers building speech-to-text features into their own products should watch whether Gemini dictation quality shifts user expectations around what "good enough" means for free. AssemblyAI and similar API providers will feel that pressure downstream if it does.