Mufal is a bot-free AI copilot that helps during live meetings—not only afterward. Get live transcription, contextual answers, an undetectable overlay during screen sharing, automatic notes, action items, and durable project memory. It works out of the box with Mufal-managed AI, while advanced users can bring their own API keys. Available on macOS, Windows, and iOS, with a free plan and optional cloud sync.
GPT-5.6 made Mufal’s model-flexible architecture truly useful for live work. Mufal lets users choose the model that fits each meeting through OpenRouter, including GPT-5.6 Sol. Sol’s frontier reasoning turns a live transcript and durable project memory into concise, contextual answers during high-stakes conversations—not after they end. Terra provides a strong balance for everyday meetings, while Luna’s speed and lower cost make continuous, low-latency assistance practical. Together, they let Mufal match intelligence, speed, and cost to the moment: Sol for complex sales, consulting, or technical discussions; Terra for daily work; and Luna for fast follow-ups and summaries. This flexibility powers a bot-free copilot that can assist live, stay out of screen-share captures, and convert every meeting into decisions, actions, and reusable project memory.
Hey Product Hunt! 👋
I built Mufal because most AI meeting tools become useful only after the conversation is over. In an important call, the hardest moment is often right now: understanding a question, recalling context, answering clearly, and still taking good notes.
Mufal is a bot-free copilot for live meetings. It provides real-time transcription and contextual answers, stays out of screen-share captures, and turns each conversation into notes, decisions, action items, and lasting project memory.
It works immediately with Mufal-managed AI—no API-key setup required. If you want more control, BYOK is available as an optional setting. Meeting content is stored locally by default, with optional cloud sync.
Mufal is available on macOS, Windows, and iOS, with a free plan to get started.
I’d love your feedback, especially on the live assistance experience and project memory. What would make an AI copilot genuinely useful in your meetings?
Report
The undetectable overlay during screen sharing is the part that would make me pause before recommending this to anyone. If it stays invisible during a screen share, that means the other people on the call have no way of knowing you have live AI assistance feeding you answers in real time. That is a very different thing from a visible notetaker bot, since at least everyone on the call knows a bot is there.
Are you leaving the disclosure decision entirely up to the user, or is there anything built in that nudges toward telling the other participants an AI copilot is running. Also curious about the contextual answers part specifically. If it is suggesting what to say back to someone in the middle of a negotiation or an interview, how do you handle it when the suggestion is wrong or based on stale context. A bad transcription glitch turning into a bad live answer seems like the failure mode that would actually matter here.
@thys_beesman That’s a fair concern. Mufal isn’t designed to help people fake expertise. It’s for people who know their subject but may struggle under pressure, during a sales presentation, an important meeting, or when communicating in a second language.
We’re also building a Projects feature where users can upload their own documents, presentations, product information, and meeting notes. Our RAG system uses these materials to ground suggestions in the user’s actual content instead of relying only on a live transcript or the model’s general knowledge. It acts more like an extended memory: the brain can’t retain every figure, detail, or document during a conversation.
Of course, transcription and AI suggestions can still be wrong. Nothing is spoken or sent automatically, the user remains in control and must decide what is relevant. We’re also working on showing uncertainty and avoiding suggestions when the available context isn’t reliable enough.
The invisible overlay is meant to keep personal notes and translations from being accidentally shared, not to encourage deception. Our goal is to help people communicate what they genuinely know with more clarity and confidence.
Mufal
Mufal
Mufal
@thys_beesman That’s a fair concern. Mufal isn’t designed to help people fake expertise. It’s for people who know their subject but may struggle under pressure, during a sales presentation, an important meeting, or when communicating in a second language.
We’re also building a Projects feature where users can upload their own documents, presentations, product information, and meeting notes. Our RAG system uses these materials to ground suggestions in the user’s actual content instead of relying only on a live transcript or the model’s general knowledge. It acts more like an extended memory: the brain can’t retain every figure, detail, or document during a conversation.
Of course, transcription and AI suggestions can still be wrong. Nothing is spoken or sent automatically, the user remains in control and must decide what is relevant. We’re also working on showing uncertainty and avoiding suggestions when the available context isn’t reliable enough.
The invisible overlay is meant to keep personal notes and translations from being accidentally shared, not to encourage deception. Our goal is to help people communicate what they genuinely know with more clarity and confidence.