LM-Kit OneThe Private AI Application Server
Promoted
Launched this week
MosMos
Voice writing that works before, during, and after meetings
842 followers
Voice writing that works before, during, and after meetings
842 followers
MosMos goes beyond voice dictation by turning both individual thoughts and group conversations into usable writing. Speak naturally in any app, get fast and accurate text in the style you choose, or ask MosMos to search the web for current information. Its personal glossary remembers specialized terms after you add them once. In multi-speaker meetings, it tracks precise timestamps, distinguishes speakers, and creates structured notes, summaries, decisions, and action items for easier follow-up.
The app specific writing is interesting. Does it feel different in Slack compared to other apps?
MosMos
@jack_mark8 Yes—that’s the idea. In Slack, MosMos keeps the writing shorter and more conversational, while emails are more complete and docs more structured. It detects the active app automatically, and you can always change the style.
MosMos
@jack_mark8 Hi, Jack! Thanks for coming!
Curious which apps you’d use this in most! Are you mainly switching between Slack and email, or would you also use voice for docs and coding prompts?
how does it manage privacy when transcribing multi speaker meetings with senstive company discussions?
MosMos
@charles_brewer90 Great question. Meeting processing currently uses cloud services. The recording is processed to create a speaker-separated transcript。
MosMos
@charles_brewer90 Just to add: we don’t retain meeting audio on our servers after processing. Our providers are also contractually required to delete the content once the request is complete and not use it for model training.
Triforce Todos
how many speakers can it actually handle before it starts mixing people up in meetings?
MosMos
@abod_rehman Great question! 2–6 people is the sweet spot from what we’ve seen. MosMos uses our in-house MOSS Transcribe Diarize Pro model. It can handle more, but similar voices background noise are usually what cause mix-ups.
MosMos
@abod_rehman Curious what your usual setup looks like—how many people, and are they sharing a room or joining remotely? If you give MosMos a try, let us know where it mixes speakers up. Those examples would be really useful for improving it.
How does speaker separation hold up when people interrupt each other or briefly talk at the same time?
MosMos
@carlvert We actually need our own ASR model (MOSS-Transcribe-Diarize) for speaker separation, so it has very top performance during interruptions and overlapping sentences. Welcome to try it out!
MosMos
@carlvert To add a little context, our own MOSS-Transcribe-Diarize model handles transcription and speaker separation together. The goal is to keep track of who’s speaking even when the conversation gets messy.
Brief interruptions are different from several people speaking over each other for a sustained stretch. That second case is less common in the meetings we’ve seen, but it’s also harder, and speaker labels or individual words can still be wrong. If you try it with your meetings, we’d really appreciate hearing where it works well and where it needs improvement.
Does the web search feature work from any app where MosMos is active?
MosMos
@yangliu44 Thanks for asking！
Yes! Press Fn+Space (or your custom shortcut) from the app you’re working in to bring up MosMos and ask it to search the web. You can also revisit the source links in History to check the original pages afterward.
Can you jump directly from a summary point back to the exact audio segment?
MosMos
@devinstone Not yet, but that’s a great suggestion!
Jumping straight from a summary point to the audio would make checking the original context much easier. We’ll explore adding this in a future update.
Thanks for the idea!！
MosMos
@devinstone @lego260312 Great suggestion!
Meeting notes are easy to generate but harder to make actionable. How does it decide what counts as a decision versus an action item?
MosMos
@bastienrenard Good question!
MosMos uses an LLM to analyze the ASR transcript, interpreting the context and intent to distinguish agreed decisions from follow-up tasks. It can still misinterpret a discussion, so we’d love for you to try it with your own meetings.
If it gets something wrong, please let us know—we’ll review the feedback and work on improvements promptly.