Rawfootagein.Scroll-stoppingshortsout.
Real reels, straight from the editor.
Not a template. An editor that makes decisions.
What to cut, where to zoom, which moment to open on, then it reviews its own cut and fixes what’s off.
Finds your hook
We transcribe your footage and open on the strongest moment, front-loaded for the scroll.
Cuts the dead air
Silences and filler words (“um”, “uh”) are trimmed to the word. Every second earns its place.
Captions that read
Word-by-word captions with keyword emphasis, built strictly from what was actually said.
Reframes to the speaker
A 9:16 crop that follows the subject, with measured push-in and punch zooms on the beats.
Beat-synced & ducked
Tempo detection snaps cuts to the beat; music ducks under speech automatically.
Edits to what it sees
A vision model reads your actual frames - favouring the sharp, well-framed takes and keeping the speaker's face centred, never half-cropped. On every plan.
Reviews its own cut
Every render gets a self-review pass that flags awkward cuts or unreadable captions and re-renders to fix them. Same on Free as on Studio.
Multiple variants
Every upload returns several editorial takes (safe, high-energy, hook-led), so you choose.
Watch a raw clip become finished reels.
The same pipeline that runs on every upload: transcribe, cut, reframe, caption, review, split.
Upload raw footage
Drop in your clip; pick a platform and editing style. Talking-head, podcast, demo. It adapts.
It analyzes and edits
Transcribe, find the hook, cut dead air, reframe, caption, grade, and review, automatically.
Pick a take and post
Platform-ready 9:16 cuts in multiple variants. Preview, adjust the timeline, download.
A style for every format.
Pick the treatment that fits your content; we adapt the cuts, captions, and pacing to match.
Same AI quality on every plan.
Every plan runs the exact same AI editor: vision-aware planning that edits to what's on screen, active-speaker auto-reframe (face always centred), word-by-word captions, beat-synced cuts, colour grade, and a self-review QC pass. Plans differ only on how much you make, how long your source can be, upload size, and the Free watermark - never on quality.
Add-on packs (Creator & Studio): minutes +30 / $9 · +100 / $29 - renders +10 / $9 · +40 / $29.
Built to make every reel scroll-stopping - punchy cuts, dynamic captions and a professional finish.
Automate it with the API and MCP.
Upload footage and fetch finished reels programmatically over a clean REST API, or let AI agents drive it through the built-in MCP server (Claude and other MCP clients). Create a key, authenticate with a bearer token, and go. Usage counts against your normal plan quota, with nothing extra to buy.
POST /api/v1/videos- create + get an upload URLPOST /api/v1/videos/{id}/finalize- start processingGET /api/v1/videos/{id}- status + output URLsGET /api/v1/videos- list your videos
# 1. Create a video (returns a presigned upload URL)
curl -X POST https://reelsmith.video/api/v1/videos \
-H "Authorization: Bearer $REELSMITH_KEY" \
-H "Content-Type: application/json" \
-d '{"title":"My clip","platform":"YOUTUBE_SHORTS",
"editingStyle":"TALKING_HEAD","filename":"in.mp4",
"contentType":"video/mp4","byteSize":52428800}'
# 2. Upload bytes -> 3. finalize -> 4. poll GET /api/v1/videos/{id}
# when status = READY, "outputs" has each take's downloadUrlA built-in Model Context Protocol server at /api/mcp. Point Claude or any MCP client at it with your key and it can create, track, and fetch videos. No server to host.
Hashed API keys, workspace-scoped access, per-key rate limiting, strict input validation, and responses that return only signed, short-lived media URLs. No secrets or storage keys are ever exposed.