Product demos · walkthroughs · bug reports

Turn recordings into
polished walkthroughs.

Record once, then let Smart Screen understand your clicks, narration, and workflow. Add polished zooms, captions, trims, and callouts without giving up control of the final edit.

Download for Apple Silicon See the workflow →
macOS arm64 installer · v1.1.3 release asset
Click-aware editing Narration-powered captions Optional BYOK AI Offline-first core flow
✦ Smart Screen applied CC captions in timeline
Polish pass
1 click to refine
Apply captions, smart zooms, trims, and click emphasis in one pass.
What it does
01 understand the recording
02 suggest the important moments
03 export a polished demo
50+ IPC channels bridging
main ↔ renderer
3 Bundled native macOS
telemetry helpers
0 Cloud dependencies
for core features
CC Dedicated captions track
seekable in timeline
2 AI providers supported
(OpenAI + Ollama)

Most screen recorders capture everything.
Great demos should focus only on what matters.

Smart Screen sits in the useful middle: it reads clicks, narration, silence, and motion to suggest a cleaner story, while still letting you choose exactly what gets applied.

Native macOS

Capture the signals that make
a demo understandable.

Three bundled native helpers run alongside your recording to capture the context normal recorders miss: real click positions, keystroke timing, and speech transcription. That means the editor can understand intent, not just pixels.

MouseClickMonitor.app KeyboardShortcutMonitor.app 10 Hz cursor sampling
🔒 Built for accurate demos, not just generic screen capture
Smart Screen launch controls with streamlined recording options
Transcription → Timeline

Narrate naturally.
Get captions that stay editable.

Record with a mic and get an editable caption track automatically. Review the transcript, fix wording, seek by segment, and burn captions into the final export without leaving the editor.

macOS native transcription SRT / VTT import CC row in timeline Click to seek
Transcript and caption editing interface inside Smart Screen
Smart Screen Analysis

Useful even without AI.
Better when you want it.

The local Smart Screen pass already gives you useful zoom, trim, and focus suggestions the moment a recording opens. AI is optional refinement, not a requirement for the product to feel smart.

Click pattern detection Silence detection Auto-zoom suggestions Before / After preview
Smart Screen options panel with smart polish controls
BYOK AI

Bring AI when it helps.
Keep control when it doesn’t.

Use OpenAI for stronger multimodal refinement, or point to a local Ollama model when privacy matters more. The app keeps secrets in the main process and lets you apply AI suggestions selectively instead of blindly accepting them.

OpenAI GPT-4o Ollama local inference Secrets in main process only Selective AI apply
AI settings and provider options inside Smart Screen
The flow

From rough recording to polished demo
in four steps.

The core workflow works locally by default. AI can sharpen the result, but the product is already useful before you add it.

🎬
01
Record

Screen + mic + native click/keystroke telemetry captured simultaneously. Saves as .webm with cursor JSON and audio sidecars alongside.

✦
02
Analyze

Smart Screen heuristics run the moment the editor opens. Auto-transcription fires if a .transcription.wav sidecar exists. No waiting, no manual trigger.

✂️
03
Polish

Review zooms, trims, and captions in the timeline. Apply AI suggestions selectively. Toggle between raw and polished preview to see the exact transformation.

📤
04
Export

MP4 or GIF with captions burned in, click emphasis, keystroke overlays, and applied effects. Share the file — not a screen recording app link.

Three operating modes

Useful on day one. Flexible when you need more.

Start with the local editing flow, then choose cloud AI or local Ollama only if your workflow benefits from it.

Offline

Local heuristics only

Full Smart Screen analysis without any network call. Works on a plane. Works in a secure environment. Works on day one with no setup.

  • Click, typing, silence detection
  • Auto-zoom and trim suggestions
  • Native macOS transcription
  • No API key, no internet
Hybrid

Local + BYOK AI refinement

Run local heuristics first, then optionally pass to OpenAI for transcript-grounded zoom planning, step titles, and callout suggestions.

  • OpenAI GPT-4o support
  • Transcript-aware AI suggestions
  • Secrets in main process, never in files
  • Selective apply — not auto-accept
Local AI

Ollama on-device

Route model calls to a local Ollama instance. Auto-discovers models, distinguishes base vs instruction-tuned, works on air-gapped machines.

  • Zero data leaves your machine
  • Auto model discovery
  • Vision-capable model detection
  • Works fully offline
Demo

Watch the full walkthrough.

From raw recording to polished export — the complete flow in under three minutes.

demo.mp4 · recorded with Smart Screen · captions, zooms, and polish flow included
Under the hood

Electron + React + TypeScript.
Deep native integration.

The main process handles recording, native telemetry, transcription, and AI provider calls. The renderer handles the UI. 50+ IPC channels connect them — keeping secrets and privileged APIs where they belong.

50+
IPC contextBridge channels
3
Native macOS helper apps
CC
Captions track in timeline editor
AI
Providers: OpenAI + Ollama
architecture.ts
// Main process — privileged
const mainProcess = {
  recording: "lifecycle + native capture",
  telemetry: "click · key · cursor",
  transcription: "MacOSTranscriber.app",
  ai: "OpenAI | Ollama",
  secrets: "electron-store (encrypted)",
};

// Renderer — React UI
const renderer = {
  editor: "timeline · zooms · captions",
  preview: "Pixi.js overlay rendering",
  export: "MP4 | GIF + burned-in CC",
};

// Bridge — 50+ channels
window.electronAPI.ipcBridge();