Launched this week
ExtraBrain
Private AI copilot for interviews and meetings
19 followers
Private AI copilot for interviews and meetings
19 followers
ExtraBrain is a free macOS AI copilot for live interviews, system design rounds, debugging calls, and technical meetings. It runs as a private overlay visible to you, can stay out of macOS screen recordings/screen sharing where supported, listens/transcribes, reads selected screen context, and gives concise real-time analysis with local transcription, BYO providers, custom endpoints, and built-in profiles.
Hey Product Hunt! I'm Dzmitry, one of the makers of ExtraBrain.
We built ExtraBrain as a private AI overlay for live technical conversations - visible to you, screen-aware, and designed to stay out of shared screens, screenshots, and recordings where supported.
The problem we kept running into: live interviews, system-design rounds, debugging calls, and meetings move faster than a normal chat window. The useful context is not only what someone just asked. It is also what is visible on screen, what was already discussed, and what kind of conversation you are in.
ExtraBrain is a macOS desktop AI copilot that can:
- transcribe live microphone and system audio
- use selected screenshot/screen context
- generate concise real-time analysis and follow-up ideas
- switch between Coding, System Design, Behavioral, Meeting, and Assistant profiles
- save local session history with transcripts, screenshots, analyses, and extracted facts
- use your own OpenAI, Anthropic, Claude, Codex, Deepgram, or OpenAI-compatible provider setup
- run local speech-to-text with NVIDIA Parakeet
- run fully locally with local speech-to-text and a local Gemma 4 model, hardware permitting
One thing we care a lot about is user control.
We do not force everyone into one hosted AI stack or bundled AI credits. Provider keys are stored locally. LLM requests go directly to the provider or endpoint you choose. If you use OpenAI, Anthropic, Deepgram, or another cloud provider, the relevant content is sent to that provider. If you want maximum privacy, you can use local transcription and a local model setup instead.
The core app is free. Pro is mainly for custom profiles and premium model options.
For Product Hunt, we're offering PHBRAIN40: 40% off ExtraBrain Pro for the first year for the first 200 users. The free core app remains free.
A few honest tradeoffs:
- macOS first today; Windows and Linux are coming later
- best results require configuring providers, local CLIs, or local models
- local models depend on your hardware
- cloud LLM/transcription providers receive the content you choose to send them
- users should follow interview, workplace, meeting, and platform rules
We'd love feedback on:
1. Is the positioning clear: interview assistant, meeting copilot, or live technical-session copilot?
2. Which profile should we improve first: Coding, System Design, Behavioral, Meeting, or Assistant?
3. What would make BYO provider setup easier?
4. Would you prefer a simpler hosted setup, or more local/offline controls?
Thanks for checking it out - would love your honest feedback.
mailX by mailwarm
Live system design and debugging rounds are incredibly high-pressure. A private, native copilot with BYO keys and local transcription—completely hidden from screen shares—is a massive security blanket for developers.
Quick question: How does it keep the overlay 100% invisible during a live Zoom or Teams call? Does it rely on native macOS window sharing APIs under the hood?
@manal_essalek1 Thanks - that’s exactly the high-pressure use case we built ExtraBrain for. On macOS, the shared content and ExtraBrain are separated at the app/window level, and we handle that through the desktop app workflow with Electron.js APIs, so ExtraBrain stays out of the shared view during screen sharing. It’s also a desktop app, not a meeting bot joining the call.
mailX by mailwarm
congrats on your launch!
@naimz Thanks