Point, speak, and Genie Lite packages your screen, your words, and exactly what you pointed at into a shareable storyboard — drop it straight into ChatGPT, Claude, Figma Make, or any AI tool. No more cropping, captioning, or re-explaining context.
Apple Silicon Mac · Microphone, Screen Recording, and Accessibility permissions · one-time speech model download of about 1.6 GB.
AI access and hosted sharing are included in the beta. No personal API key or storage setup needed.
How it works
Genie Lite captures everything an AI tool needs from your screen and voice — and hands you a storyboard to paste in.
Press Cmd+G and say what you need in English or Chinese while you point at what matters on screen.
Cmd + GGenie grabs your screen, reads it, and marks exactly where your cursor was — turning your moment into clean, first-person context.
Get a storyboard — your words plus annotated screenshots. Copy the image or share a link, then paste it into ChatGPT, Claude, Figma Make, or any AI tool.
Cursor-aware
While you talk, Genie Lite follows your cursor and traces the path you moved. Where you land becomes a red marker on the screenshot — so "make this rounder" or "fix this bug" points to the exact spot, no description needed.
"Make this button rounder." → the marker shows the AI exactly which one.
Demos
A few real captures — from pointing on screen to a storyboard pasted into an AI tool.
Features
Built with native macOS APIs for speed and privacy — and designed to work with whatever AI tool you already use.
Local speech-to-text via Whisper. No internet required for transcription. Just speak naturally and Genie listens.
Captures screenshots, reads text via OCR, and tracks your cursor — then drops a red marker on the exact spot you pointed to, so the AI knows what "this" means.
The storyboard is just text and images — so it drops into ChatGPT, Claude, Figma Make, Claude Code, or anything else. Copy it as one image, or share a hosted link.
Copy the whole storyboard as a single image for chat tools that read pasted pictures — or grab a hosted link for agents that fetch URLs. Whatever your tool prefers.
Built on Tauri with Rust. ~32 MB binary, ~30 MB idle RAM. Direct macOS API access — no Electron, no bloat.
Use cases
Capture the moment, hand the storyboard to your favorite AI tool, and skip the screenshotting and explaining.
Point, speak, and share the context. Enable macOS capture permissions before your first recording; AI access is included.
Current download: v0.1.13 beta, signed and notarized by Apple. Includes automatic storyboard image copying, more reliable screenshots, and hosted links without API-key setup. Automated checks passed; testing on a fresh macOS account is still outstanding.
Apple Silicon (M1/M2/M3/M4). Requires macOS 12.3+. Intel build coming soon.
Allow Microphone for speech, Screen Recording for screenshots, and Accessibility for window and document context. AI access is included; no personal API key is needed. Let the ~1.6 GB speech model finish downloading, and restart if macOS asks.
Point and speak while the relevant content is visible. The complete storyboard image copies automatically; press Cmd+V in your AI chat to paste it. Use Copy link if you prefer a hosted link. No storage setup is needed.
Speech transcription runs on your Mac. Visual review sends screenshots and your transcript to Claude. Recent-activity summaries can also send background screen text. Hosted storyboard links are accessible to anyone with the link.
Changelog
v0.1.13 is available from the download above and through the automatic updater. It restores automatic image copying and includes the screenshot fixes from v0.1.12. The installer and updater app are signed, notarized, and stapled, and pass Apple's verification. macOS can still show its standard first-open and permission prompts.
205 backend tests, 25 interface tests, and 16 service tests passed, including automatic image-copy command routing, explicit link copying, retries, and stale-storyboard handling. Fresh-account native onboarding remains untested.
Release verification and checksums · Versioned v0.1.13 download.
These screenshot fixes are included in v0.1.13, which changes the default clipboard output back to an image. The original v0.1.12 installer is retained unchanged for reference and is no longer the candidate to notarize.
205 backend tests, 22 interface tests, and 16 service tests passed. Replaying the reported failed recording recovered 10 screenshots with cursor trails. Whole-request AI failures can still prevent visual review; fresh-account native onboarding remains untested.
Archived v0.1.12 handoff. These fixes are included in the released v0.1.13 above.
The original v0.1.11 installer is retained unchanged for reference. The released v0.1.13 keeps included AI and hosted links but automatically copies the storyboard image instead of the link.
201 backend tests, 22 interface tests, 16 service tests, and a replay of the reported screenshot failure passed. Fresh-account native onboarding remains untested. Genie's shared service no longer imposes daily, cumulative, or request-frequency quotas; provider availability and per-request technical constraints still apply. Anyone with a storyboard link can view its contents.
v0.1.10 was the previous notarized release and has been replaced by v0.1.13 in the default download and automatic updater. Its Chinese support and widget fixes remain included.
195 Rust tests, 18 interface tests, and a local Mandarin audio test passed. Chinese image/PDF exports were checked. Fresh-account onboarding and further native recording/widget testing remain outstanding.
These v0.1.9 improvements are also included in v0.1.10. Apple notarization and automated checks passed; a fresh-account onboarding test is still outstanding.