Verismo — native screen studio for macOS

Features

Everything between “stop recording” and “looks great”

The same renderer drives the editor preview and the export — what you see is exactly what you ship.

Designed framing

Pad, round, and float the capture on a solid, gradient, blurred, image, or animated procedural backdrop — grid-pulse, dots, waves.

Auto-zoom from clicks

Your click clusters become smoothed zoom keyframes. Hand-edit any of them, or regenerate the whole envelope with one button.

A cursor worth watching

A synthetic cursor — dot, ring, arrow, or square — with halo glow and click ripples, smoothed over the raw event stream.

Webcam, two ways

Round picture-in-picture in any corner — or a person cutout keyed straight onto your screen content. No green screen needed.

Annotations & music

Text, arrows, and highlights scoped to the timeline; an MP3/M4A music bed with trim and volume handles. Both bake into the export.

Transcripts & captions

One click transcribes your narration, Refine with AI cleans it up, and Apply as Captions drops it onto the video. Uses your own OpenAI key.

Speaker notes

A floating teleprompter you read while recording — auto-scroll or drag. It’s excluded from capture, so only you ever see it.

A library that ships

Every take in one searchable grid. Multi-select for bulk delete or a queued Export N — and import .mov/.mp4 you recorded elsewhere (⇧⌘I).

Usage

One take, three moves

A recording session in Verismo reads like its own timeline.

Record

Pick a display, arm the mic and camera, and hit the red button on the floating control bar. The bar stays above your windows but out of your recording — and so does the teleprompter you read from. Clicks and cursor moves are logged alongside the video for the edit.

Polish

The take opens in the editor. Generate auto-zoom from your clicks, style the cursor, choose a backdrop, add captions, annotations, and a music bed. Every control renders live — the preview is the export.

Export

One button renders a single MP4 that matches the preview pixel-for-pixel. Shipping a batch? Select five takes in the Library and export them as a queue with a shared preset.

for agents — the verismo CLI, JSON in / JSON out

$ verismo capabilities | jq '.commands[].name'   # discover the surface
$ verismo record start --mic --json              # begin a take
$ verismo record stop --json                     # → bundle path
$ verismo export "$BUNDLE" --preset mp4-1080p30  # render the movie

Screenshots

The studio, at a glance

Straight captures of Verismo 0.11.3 — no mockups. The editor itself is up top, framed the way Verismo frames your recordings.

The Verismo Library: a searchable grid of recordings with duration badges, an Import button, and a green New Recording button. A sidebar lists General, Themes, Recording, Shortcuts, AI, and About settings. — **The Library.** Every .osmoproj in one grid — search, sort, multi-select, bulk export. Settings for themes, shortcuts, recording, and AI live in the sidebar.

The floating control bar in its compact Ready state. — **The control bar.** The whole app while you record — and it never appears in the take.

Privacy

Runs on your Mac. Stays on your Mac.

Verismo has no server, no account, and no analytics. There is exactly one situation where data leaves your machine — and you hold the key to it.

Local by default

Recordings are plain .osmoproj bundles in ~/Movies/Osmo Recordings — video, mic audio, cursor log, edits. Open, back up, or delete them in Finder.
Recording, editing, rendering, and export all run on-device. Nothing is uploaded, ever.
No account, no telemetry, no crash reporting, no update phone-home. The app is MIT-licensed — you can read every line.

AI features: bring your own key

Transcripts, Refine with AI, captions, and Smart Trim stay off until you paste your own OpenAI API key in Settings → AI.
When you trigger them, your narration audio (transcription) or transcript text (refine, trim suggestions) goes directly from your Mac to api.openai.com — there’s no middleman.
That data is then handled under OpenAI’s privacy policy, on your own OpenAI account and billing.
The key lives in a file only your user account can read, and you can remove it in Settings → AI at any time. No key, no traffic.

SCREEN RECORDINGTo capture the display you choose. macOS asks the first time.

MICROPHONEOnly if you narrate. Skip it and record silent takes.

CAMERAOnly for the webcam overlay or person cutout.

INPUT MONITORINGLogs clicks and cursor moves that power auto-zoom and the smoothed cursor. The log stays in the project bundle.

FAQ

Before you ask

Short answers to the usual questions.

How much does Verismo cost?

Nothing. Verismo is free and MIT-licensed — no purchase, no subscription, no account.

Which Macs does it run on?

Any Mac running macOS 14 (Sonoma) or later.

Do I need an OpenAI API key?

No. Recording, editing, and export are fully local and never need a network connection. A key only unlocks the optional AI features — transcripts, Refine with AI, captions, and Smart Trim — and you add it yourself in Settings → AI.

Where do my recordings live?

In ~/Movies/Osmo Recordings, as self-contained .osmoproj bundles holding the video, mic audio, cursor log, and your edits. The Library window lists them all; Finder works too.

Can I polish a video I recorded somewhere else?

Yes. Import a QuickTime .mov or .mp4 (Library → Import, or ⇧⌘I) and Verismo turns it into an editable project — framing, zoom, cursorless captions, music, and all.

What exactly does export produce?

A single MP4 that matches the editor preview pixel-for-pixel — same renderer, same frame. Presets cover common sizes (e.g. 1080p30), and the Library can run a queued export across many takes.

Can AI agents drive Verismo?

Yes — the bundled verismo CLI speaks JSON in and out, auto-launches the app headless, and covers recording, editing, and export. Claude Code and Codex can use it as a tool.

Record the screen. Ship the movie.