Ellis is a consumer-first AI notetaker designed specifically for in-person meetings. It empowers individuals to capture conversations from various real-world scenarios, such as coffee meetups, on-site sales meetings, therapy sessions, doctor visits, interviews, and parent-teacher conferences. The primary purpose of Ellis is to provide users with accurate transcripts and the ability to query their recorded conversations, offering insights and recall without the need for a laptop or additional hardware, relying solely on an iPhone or Apple Watch.
The problem Ellis addresses is the difficulty of accurately capturing and recalling information from in-person discussions, especially when compared to the more structured environment of online meetings. Traditional note-taking can be time-consuming and prone to errors, leading to missed details or decisions. Ellis aims to solve this by providing an automated, intelligent solution that not only transcribes conversations but also identifies speakers and allows for natural language querying of the content, ensuring that valuable information is not lost.
A key feature of Ellis is its ability to accurately identify speakers in a real-world setting, which is more challenging than in online calls where each voice is isolated. Ellis achieves this through voice enrollment and sophisticated diarization techniques, coupled with a user-friendly interface that allows for quick tagging of oneself and other participants. This ensures that transcripts clearly attribute statements to the correct individuals, enhancing the clarity and usability of the notes.
Another significant capability is the ability to ask questions about any recorded conversation. Users can retrieve specific information, recall decisions made, or revisit any detail discussed simply by asking. This interactive feature transforms static transcripts into dynamic knowledge bases, allowing for efficient information retrieval and review.
Ellis also offers a unique way to find information by location. If a user forgets a name or a specific detail, they can query the system based on where the conversation took place, such as asking, "what did we agree on during our walk in Fort Greene?" This contextual search functionality adds another layer of accessibility to the recorded data.
Privacy is a core consideration for Ellis. The product is designed to be private by default, with recordings automatically deleted once the transcription process is complete. This ensures that sensitive information captured during meetings is handled with care and not retained unnecessarily.
Ellis operates by recording audio from in-person meetings using an iPhone or Apple Watch. The audio is then processed to generate a clean transcript with speaker identification. Users can then interact with the transcript by asking questions to retrieve specific information or details from the conversation. The system uses voice enrollment to match the user's voice and provides tools to tag other speakers, ensuring accurate attribution.
The benefits for users include improved recall of meeting details, better decision-making through accurate record-keeping, and the ability to synthesize information across multiple conversations, both professional and personal. By centralizing conversational data, Ellis allows users to identify trends and commonalities across different aspects of their lives.
Concrete use cases for Ellis include capturing notes during client sales meetings to track agreements and follow-ups, recording therapy sessions to review therapeutic insights, documenting doctor's visits to remember medical advice and treatment plans, and taking notes during job interviews to recall key questions and impressions. It can also be used for personal conversations where remembering details is important.
Ellis is designed for individual users and is available on iPhone and Apple Watch. The product is free to use. It leverages technologies like AssemblyAI for speaker diarization and Pyannote for speaker embeddings, with a UI that allows users to explicitly select and label speakers. The company also mentions using Whisper by OpenAI and Supabase.
In summary, Ellis provides a seamless and private way to capture, transcribe, and query in-person conversations, transforming spoken interactions into actionable insights accessible directly from a user's iPhone or Apple Watch.
