Real-time stealth meeting copilot for Windows with local ASR and BYOK LLM
MeetingCopilot is a Windows application that provides real-time transcription of the other party in meetings or interviews, using either local FunASR or cloud ASR. It displays suggested answers on a first-person teleprompter that is invisible to screen sharing, powered by a bring-your-own-key (BYOK) LLM. It solves the problem of needing discreet, real-time assistance during remote meetings without being detected.
Visit project ↗Entities
Related
EarWitness: local-first meeting transcription with speaker diarization
EarWitness is a locally hosted desktop transcription app that taps your audio device to transcribe, segment, and diarize speakers in meetings, all on your machine. It uses LLMs (Claude via CodeMySpec) to build the app and likely for transcription/diarization. It solves the problem of needing a meeting bot in participant lists by recording directly from your audio device.
Notch: invisible voice-following teleprompter for meetings
Notch is a desktop overlay that displays notes invisibly during screen sharing and automatically scrolls to follow the user's speech using a local voice activity model. It helps professionals in meetings, interviews, or presentations stay on script without distracting others.
Open-source private desktop AI overlay with MCP support
Wisp is a desktop overlay that lets you invoke AI from any app via keyboard shortcuts, with context like text, screenshots, or active window. It supports voice queries, dictation, and pasting results directly, keeping data local and private.
AI-powered desktop overlay for coding assessments and interviews
HintLens is an AI-powered desktop overlay that stays on top of the screen to provide answers during coding online assessments and interviews. It uses LLMs to help users avoid switching between apps or browser tabs, reducing stress from strict time limits and challenging questions.
CouncilAI: local desktop app routing questions to 4 AI models
CouncilAI is a Windows desktop app that runs four local AI models and routes user questions to the appropriate model based on complexity. Simple questions use a fast lightweight model, while complex ones trigger deliberation where multiple models answer and the best response is selected. It operates fully offline on the user's hardware, requiring no accounts or cloud services.