ElevenLabs, the voice-AI company valued at $22 billion, is entering the crowded market for meeting notes and dictation with an experimental app called ElevenAI, unveiled at its 11/11 Summit in New York.
The company is positioning ElevenAI as a product for teams and enterprises rather than an individual productivity tool — a quieter way of competing with the standalone notetaker licences companies buy today. The app, still in preview and available only through an early-access sign-up, records and summarises meetings, notifies other participants that a conversation is being recorded, and lets users query their meeting history — including live assistance during a call to recall a detail from an earlier conversation. Follow-up work can be delegated to ElevenAgent: drafting a reply from a meeting's notes, for instance.
ElevenAI also does plain dictation, converting speech into cleaned-up text in any application and supporting more than 90 languages. The company says it understands industry-specific vocabulary in both dictation and transcription — the same differentiator its rivals are pushing.
The launch puts ElevenLabs directly alongside Fireflies, Wispr Flow and Granola, all of which have shipped competing capabilities in the past few months. ElevenLabs' advantage is that it owns the underlying speech models, so ElevenAI doubles as a showcase for its own technology and a funnel for its agent platform. Its pitch to enterprises is consolidation: fewer licences, one vendor, and meeting history held in a place that can already act on it.
In the same week, ElevenLabs released its v3 series of Scribe speech-to-text models, adding live transcription, fine-tuning on a customer's own data, a tool that picks the right language model for a task, and a routine creator that automates actions from a transcript.
Whether teams want their note-taker to also be their agent platform is the open question. The category is converging on the same promise — becoming the place where a company's spoken knowledge is stored and acted upon — and the winner will likely be decided less by transcription accuracy than by which vendor teams already trust with the rest of their workflow.

