Open-source alternative to Cluely. A stealth meeting-assistant overlay invisible to screen-share, with agent-driven answers wired to your model. Speech-to-speech pipeline, local-first. Run Ollama on-device for zero egress, or OpenRouter for frontier models. Five built-in agents (interview-coding, system-design, behavioral, sales-discovery, notetaker) - each a TOML file you can edit and share. FastAPI daemon, OpenAPI docs. Apache-2.0. No signup, no card.

Framer — Launch websites with enterprise needs at startup speeds.

Launch websites with enterprise needs at startup speeds.

Hey everyone - I built cue because the closed-source meeting overlay (Cluely) was a great UX wrapped around lock-in: their model, their pipeline, their data. cue is the open-source counterpart. Same stealth-overlay UX (invisible to screen-share/record), but every layer is yours: Ollama + on-device STT for full local mode, OpenRouter for frontier quality, or any mix. Five built-in agents - interview-coding, system-design, behavioral, sales-discovery, meeting-notetaker - each is a TOML file you can read, edit, share. There's a FastAPI daemon at 127.0.0.1:7821 with OpenAPI docs; the GUI calls it, the CLI calls it, you can call it. Apache-2.0, no signup, no card. macOS DMG and pip install both work today; Windows + Linux builds are on the releases page. Would love feedback on the agent format and which model defaults feel best for each task. Happy to answer anything about the stealth window implementation, the speech-to-speech pipeline, or the local-vs-cloud privacy posture.

Framer — Launch websites with enterprise needs at startup speeds.

Launch websites with enterprise needs at startup speeds.

cue

Open-source meeting assistant - your model, your data

Open-source meeting assistant - your model, your data