Honest comparison
Skilly vs Gemini Live
Google's Gemini Live is a genuinely impressive multimodal model running in the browser. Skilly wraps a similar loop into a native Mac app with a cursor that actually moves. If you are learning a Mac app, that last part is the difference.
You are inside a Mac app and need help without switching windows to a browser. You want the cursor to physically move to the button you need to click, not just read instructions.
You want a general-purpose multimodal AI you can use on any device (Android, iOS, web). You are comfortable working in a browser tab and do not need a Mac-native UI.
Feature by feature
| Feature | Skilly | Gemini Live |
|---|---|---|
| Where it runs | Native macOS menu bar app — always one shortcut away | Browser tab in Google AI Studio (aistudio.google.com) |
| Trigger | Voice-gated: push-to-talk or Live Tutor mode. Nothing captured between voice events. | Start a session, continuous screen + mic during the session |
| Cursor guidance | Cursor physically moves to the exact UI element on your screen | Text and voice answers only — no cursor control |
| Skills / curriculum | Per-app Markdown curriculum (Blender, Figma, Xcode, Photoshop…) — stays current independent of model | General-purpose, no per-app context preloaded |
| Model | Single-call OpenAI Realtime API (voice-to-voice) | Google Gemini 2.x (Flash / Pro depending on tier) |
| Tab-switching | Stays out of the way in the menu bar — you stay in the app you are learning | You are in the browser, not the app — you switch windows to act on answers |
| Language support | 20+ languages auto-detected | 40+ languages (broader reach) |
| Privacy posture | Capture gated by voice activity. Screen Recording permission revokable anytime. | Screen + audio streamed to Google during session. Bound by Google Workspace terms. |
| Offline | Requires OpenAI API access but the app works without a persistent browser tab | Requires a browser session + Google account |
| Open source | Yes — Apache-2.0 (fork of farzaa/clicky, MIT) | No — Google proprietary |
| Pricing | 15 minutes free, then $19/month for 3 hours of tutoring | Free tier with rate limits, paid tiers via Google AI Studio / API |
Common questions
Is Gemini Live free and Skilly paid?
Yes. Gemini Live is free with any Google account. Skilly is $19 per month after a 15-minute free trial (no credit card to start). Skilly is also open source under Apache-2.0 — you can run it free with your own OpenAI API key (BYOK option, in beta) — but the hosted convenience tier is paid.
Why pay for Skilly when Gemini Live is free?
Three reasons users pay: (1) Skilly physically moves your cursor to the exact UI element you need — Gemini Live just talks. (2) Skilly is a system-wide menu-bar app — no Chrome tab, no Google account, works in any Mac app including offline-first tools. (3) Skilly ships with 5 free skill curricula (Blender, Figma, AE, DaVinci, Premiere) for guided learning. If you don't need any of those, Gemini Live is genuinely fine.
Can Gemini Live see my Mac apps the way Skilly does?
Yes if you screen-share to it via the browser, but the experience is different. Skilly uses macOS ScreenCaptureKit natively — instant capture, multi-monitor aware, no browser permission dance. Gemini Live works through Chrome's screen-sharing API, which is fine for casual use but adds friction every session and lacks the cursor-moving guidance Skilly provides.
Does Skilly use Gemini under the hood?
No. Skilly uses the OpenAI Realtime API for voice + vision in a single round-trip call (sub-second latency). Gemini Live uses Google's Gemini multimodal model. Different model providers, different latency profiles, different language coverage. Skilly's source code is open at github.com/tryskilly/skilly if you want to verify.
Want a Mac-native tutor?
15 minutes free, no card. Apache-2.0 open source, fork of Farza's Clicky (MIT).
Download Skilly