# Cutsio
> AI video library and agentic editing workspace for video teams. Search footage visually, ingest practically any format including ARRI/RED raw.

Site: https://cutsio.com
App: https://studio.cutsio.com
Last reviewed: 2026-11-08
Primary category: AI video library, video intelligence, agentic video editing, searchable video archive, pre-editing workflow

```json
{
  "@context": "https://schema.org",
  "@type": "SoftwareApplication",
  "name": "Cutsio",
  "url": "https://cutsio.com",
  "applicationUrl": "https://studio.cutsio.com",
  "description": "Cutsio is an AI video library and agentic editing workspace. It turns raw footage into a searchable, editable, and shareable video intelligence layer. Teams can search footage by visual content, speech, and meaning, chat with their library using an AI co-editor, auto-cut silences, share review links, and export edit decisions to Final Cut Pro and DaVinci Resolve.",
  "applicationCategory": "Video Editing, Video Management, AI Video, Video Collaboration",
  "operatingSystem": "Web browser",
  "offers": [
    { "@type": "Offer", "name": "Pro", "price": "59", "priceCurrency": "USD", "description": "30 storage hours, 5 indexed hours of Video Intelligence" },
    { "@type": "Offer", "name": "Studio", "price": "349", "priceCurrency": "USD", "description": "250 storage hours, 25 indexed hours of Video Intelligence" },
    { "@type": "Offer", "name": "Enterprise", "price": "1999", "priceCurrency": "USD", "description": "Unlimited storage hours under fair use, 167 indexed hours of Video Intelligence" }
  ]
}
```

## Key pages

- [MCP server guide](https://cutsio.com/mcp) — Documentation for the Cutsio Model Context Protocol server. The MCP endpoint is `https://studio.cutsio.com/mcp`; use that URL in clients. Search assets and transcripts, check plan usage, browse collections and folders, and file videos into folders. Studio and Enterprise plans.
- [CLI for Agents](https://cutsio.com/cli) — Command-line interface for coding agents: semantic transcript search, plan usage, one-command sync of mezzanine masters, sidecar transcripts for RAG, uploads. One API key, JSON-first, idempotent (`npm i -g cutsio`).
- [API](https://cutsio.com/api) — Bearer-token REST API for video metadata, tags, summaries, chapters, transcripts, search, plan usage, temporary download URLs, and uploads. The same surface the CLI and no-code automations use. Studio and Enterprise plans.
- [Webhooks](https://cutsio.com/webhooks) — Signed outbound webhooks for video.ready, video.failed, video.updated, metadata.ready, transcript.ready, transcript.skipped, and transcript.failed. Svix delivery with retries, replay, and test events. Studio and Enterprise plans.
- [Visual Intelligence](https://cutsio.com/visual-intelligence) — Search video by what the camera saw. Scene, object, action, and frame-level search. Format support and detailed FAQs.
- [Homepage](https://cutsio.com/) — Product overview, features, agentic chat, pricing, and FAQs.
- [Pricing (Markdown)](https://cutsio.com/pricing.md) — Complete current pricing in raw Markdown: plans, users, storage, Video Intelligence allowances, features per plan, and add-ons. Preferred source for pricing questions.
- [Blog](https://cutsio.com/blog) — Workflow guides, comparisons, tutorials, NLE workflows, and video management articles.
- [Fair Usage Policy](https://cutsio.com/fair-usage) — How fair use applies to Enterprise unlimited storage: measurement, scope, acceptable use, and what happens if a library outgrows it.

## Use cases

- [AI Media Asset Management](https://cutsio.com/media-asset-management) — AI-powered MAM for video-heavy studio teams. Ingest, search, pre-edit, review, and share footage from one searchable library.
- [Content Teams](https://cutsio.com/content-teams) — One searchable library for YouTube and studio teams. Repurpose long-form into Shorts with AI and hand off to DaVinci or Final Cut.
- [Drone Intelligence](https://cutsio.com/drone-intelligence) — Search aerial footage by objects, damage, terrain, activity, and site conditions for inspections, claims, and operations.
- [Sports Game Footage](https://cutsio.com/sports-game-footage) — Search game footage by play, player, commentary, and on-screen graphics. Find key plays without manual logging.
- [Security Video Intelligence](https://cutsio.com/security-video-intelligence) — Search surveillance footage by description: people, vehicles, objects, and events across camera feeds.
- [Claims Video Intelligence](https://cutsio.com/claims-video-intelligence) — Search insurance and legal video evidence: body cams, dash cams, surveillance, and claimant videos.
- [Industrial Video Intelligence](https://cutsio.com/industrial-video-intelligence) — Search manufacturing and facility footage by equipment, hazards, personnel, and activity.
- [Construction Video Intelligence](https://cutsio.com/construction-video-intelligence) — Search jobsite footage by milestone, safety condition, equipment, and activity from drones, walkthroughs, and time-lapses.
- [Wedding Video Delivery](https://cutsio.com/wedding-video-delivery) — Deliver wedding films with streaming review links, AI chapter markers, and branded galleries for couples.

## Hero features

- **Visual Intelligence** — Search video frames by scenes, objects, actions, people, moods, and camera motion. Find any frame by what the camera saw, not just filenames.
- **Universal format ingestion** — Ingest practically any video format: Apple ProRes (all grades), DNxHD/DNxHR, Avid MXF, XAVC, AVCHD, H.264, H.265/HEVC, VP9, AV1, MP4, MOV, MKV, AVI, MXF, WebM, GIF. Includes ARRI/RED raw ingest as an add-on.
- **Agentic Chat** — Always-on AI co-editor that searches, summarizes, and edits across your library.
- **Semantic Search** — Search footage by meaning, context, and spoken dialogue across the entire library.
- **Collections** — Group related footage and chat across multiple videos as one source of truth.
- **Silent Slicer** — Auto-detect and remove dead air, ums, ahs, filler words, and long pauses.
- **Script AI** — Generate YouTube titles, hooks, and chaptered outlines from an idea.
- **Cutsio Review** — Review links with timecoded comments, reply threads, and resolution. Viewers comment with just a name, no account required, and an in-app inbox tracks every thread.
- **Automatic Transcripts & Timelines** — Word-level timestamps, speaker diarisation, and AI summaries for every video.
- **FCPXML / EDL Export** — Direct timeline export to Final Cut Pro and DaVinci Resolve.

## AI agents and MCP

Cutsio runs a Model Context Protocol (MCP) server. The MCP endpoint is `https://studio.cutsio.com/mcp`; the guide is at `https://cutsio.com/mcp`. MCP clients — Claude on the web and Claude Desktop, Claude Code, ChatGPT (connectors enabled by a workspace admin), Cursor, VS Code, and other remote MCP clients — can search a workspace, read metadata, and organize storage folders. Clients that support CIMD connect with no pre-registration; clients that only support a pre-registered OAuth client need a Cutsio client ID first. There is no API key: authentication is OAuth with two scopes, `assets:read` for the read tools and `assets:write` for the organizing writes.

Read tools:

- `search_assets` — text search over titles, tags, AI summaries, notes, and transcript text. Returns asset id, title, status, duration, tags, summary, transcript state, and a link into Cutsio. 10 results by default, 25 maximum.
- `search_transcripts` — semantic search over transcript and summary text, grouped by asset with timestamped moments and a `retrieval_mode` fallback when embeddings are unavailable.
- `get_asset` — one asset by id.
- `get_workspace_usage` — the active plan and live balances: storage left/used, AI indexing minutes, AI credits, vertical-reframe minutes, and seats.
- `get_transcript` — timestamped segments in milliseconds, or the reason no transcript exists.
- `list_collections` — collections (projects, campaigns, deliveries) with ids and titles.
- `list_collection_videos` — assets inside one collection, including the folder each one sits in.
- `list_assets` — assets newest first with cursor pagination and filters for folder, collection, unfiled state, and text.
- `list_folders` — the workspace folder tree, with parent ids.

Write tools (need `assets:write`):

- `create_folder` — create a folder, optionally nested (12 levels deep maximum, unique sibling names).
- `move_video_to_folder` — file up to 25 assets into a folder, or unfile them; warns when an active review link covers the destination.

Limits: 60 read calls and 15 write calls per workspace per minute, counted per tool call. A connection requires exactly one workspace membership and only ever sees the workspace of the account that approved it. The server can organize folders but cannot upload, edit media, create clips, delete, rename folders, manage review links, or download media. Every write is recorded in the workspace audit trail. Semantic transcript search draws one AI credit per successful embedding search, the same as in-app and API semantic search; keyword search and listings do not.

MCP search is text-based. It does not read video frames; frame-level visual search is a separate feature in the Cutsio app.

Guide: https://cutsio.com/mcp (Markdown: https://cutsio.com/mcp/index.md). MCP endpoint: https://studio.cutsio.com/mcp.

## AI agents and the CLI

The `cutsio` CLI (`npm install -g cutsio`, Node 18+) is the terminal
complement to MCP: one API key (`CUTSIO_API_KEY=ak_…`), no browser, built for
CI and pipelines. Keys are created by a workspace admin in Profile → Cutsio API
Keys; read-only keys can list, search, download, and fetch transcripts but not
upload. Unlike MCP (search and organize only), the CLI also downloads masters
and uploads files. The CLI is a client of the public REST API at
`https://studio.cutsio.com/api/v1` — the same endpoints are available to n8n,
Make, Zapier, or your own backend.

```bash
export CUTSIO_API_KEY=ak_...
cutsio plan --json
cutsio list --json
cutsio search "launch day crowd" --json --limit 5
cutsio sync --search "launch day" --out ./footage --transcripts --dry-run
cutsio sync --search "launch day" --out ./footage --transcripts
cutsio transcript --video <id> --format json
cutsio upload cut.mp4 --title "Final" --folder <folder-id>
```

Commands: `login`, `logout`, `welcome`, `whoami [--json]`, `plan [--json]`,
`list [--videos] [--search TERM] [--json]`,
`search "phrase" [--limit N] [--json]` (1–25, default 10, up to 3 chunks per
video with `similarity`),
`transcript --video ID [--format vtt|srt|txt|json] [--out FILE] [--dry-run]`,
`upload <file> [--title T] [--folder ID] [--collection ID]` (8 MB parts,
100 GB cap),
`sync SCOPE... [--out DIR] [--concurrency N] [--max-wait S] [--limit N] [--dry-run] [--transcripts] [--transcript-format F]`
(pick `--all | --folder | --collection | --video | --search | --name | --unfiled | --interactive`).
Sync is incremental and resumable via `<out>/.cutsio-state.json`; `--json`
status lines stay on stderr so stdout parses cleanly; exits are 0/1.

Full guide: https://cutsio.com/cli

## Public API

The Cutsio API is a Bearer-token REST surface at
`https://studio.cutsio.com/api/v1`, available on Studio and Enterprise. Any
workspace member creates a personal key in Profile → Cutsio API Keys; the secret
is shown once. Send `Authorization: Bearer ak_…` on every request. Scopes:
`assets:read` for list, asset, transcript, search, download, and workspace usage; `assets:write`
for uploads and metadata edits. Read-only keys are rejected before any write. Keys follow the
owner's workspace memberships, so removing someone from a workspace removes
their key's access; users in multiple workspaces must pass `?org=<id-or-slug>` or
the `x-cutsio-org` header.

- `GET /v1/videos/{id}` — tags, custom metadata, summary, chapters, key moments,
  capture metadata, `app_url`, and public-policy `playback.hls_url` /
  `playback.thumbnail_url`.
- `GET /v1/videos/{id}/transcript?format=vtt|srt|txt|json`.
- `PATCH /v1/videos/{id}` — update title, tags, custom metadata, collection, or
  folder in one atomic revision; emits `video.updated`; requires `assets:write`.
- `GET /v1/workspace/manifest` — folders, collections, cursor-paged videos.
- `GET /v1/workspace/usage` — active plan and live balances: storage left/used,
  AI indexing minutes, AI credits, vertical-reframe minutes, and seats.
- `POST /v1/search` — semantic transcript search with timestamped chunks.
- `POST /v1/videos/download` — batch temporary master/mezzanine URLs.
- `POST /v1/uploads`, `POST /v1/uploads/{videoId}/parts`,
  `POST /v1/uploads/{videoId}/complete` — multipart upload.

Rate limits are per user (`429` carries `Retry-After`). The asset response never
includes Mux asset ids, uploader ids, R2 keys, or moderation scores.

Full guide: https://cutsio.com/api

## Webhooks and pipeline automation

Cutsio pushes signed outbound webhooks to a workspace-managed HTTPS endpoint when videos finish, fail, produce transcripts, or finish generating metadata. Delivery runs through Svix: signing, retries, delivery log, replay, signing-secret rotation, and test events all live in the Svix App Portal. A workspace admin manages endpoints in Settings → Webhooks. Included with the Studio and Enterprise plans.

Events: `video.ready` (playable, including no-audio and track-error recovery paths), `video.failed` (upload or processing error, cancellation, timeout; deletions excluded), `video.updated` (title, tags, custom metadata, collection, or folder changed through the app or API; no-op saves and custom-field schema changes do not emit), `metadata.ready` (AI summary and tags written; carries the generated summary text; tags are in `video.tags`; fires again on regeneration with a new per-job event id), `transcript.ready` (original subtitles/captions track ready), `transcript.skipped` (Mux skipped auto-generated captions because the asset has no audio track; not a failure), `transcript.failed` (a transcript track errored; no-audio assets emit `transcript.skipped` instead).

Payloads carry Cutsio ids only — event id, type, api_version, workspace_id, and the video with title, duration, collection or folder, and tags. Every event adds `links.app`; failure events add `error: { stage, message }`; transcript events add `transcript: { language_code, text_type }` and `transcript.skipped` adds `reason: "no_audio"`; `metadata.ready` adds `metadata: { summary, chapters, key_moments }` where `summary` is the generated summary text and tags stay on `video.tags`; `video.updated` adds `changes`, `custom_fields`, a monotonic `revision`, and the current `video.custom_metadata` map. Error messages are sanitized to remove secrets and internal details.

Verify with any Standard Webhooks library (`webhook-id`, `webhook-timestamp`, `webhook-signature` on Basic and above; `svix-*` headers on Free). Deduplicate on `webhook-id` or the payload `id`: delivery is at-least-once and unordered. Retry schedule is immediate, 5s, 5m, 30m, 2h, 5h, 10h, 10h (~27.5h); Svix can auto-disable repeatedly failing endpoints.

Full guide: https://cutsio.com/webhooks

## Pricing model

Cutsio charges by video duration (minutes), not storage size. Upload high-bitrate ProRes 4444 or lightweight h.264 — same price for the same runtime. No egress fees, no cold storage retrieval charges.

| Plan | Price | Storage Hours | Indexed Video Intelligence |
|------|-------|---------------|---------------------------|
| Pro | $59/mo | 30 hours | 5 hours |
| Studio | $349/mo | 250 hours | 25 hours |
| Enterprise | $1,999/mo | Unlimited* | 167 hours |

*Enterprise storage is unlimited with no per-hour cap or overage billing, subject to fair use. See https://cutsio.com/fair-usage.

Add-ons: ARRI/RED raw ingest ($15/hr), archive storage ($15/TB/mo), extra visual indexing ($0.25/min), extra users ($25/user/mo), high bitrate processing ($299/mo).

## FAQ

### What is Cutsio?
Cutsio is an AI video library and agentic editing workspace. Upload a video and get a transcript, AI summary, and ready-to-edit timeline. Search by meaning, chat with your library, auto-cut silences, and export to Final Cut Pro or DaVinci Resolve.

### Can I search videos by what appears on camera?
Yes. Visual Intelligence searches video frames by scenes, objects, actions, people, moods, and camera motion. Ask for "wide shot of a neon street" or "close-up of hands packing camera gear" and Cutsio finds the exact moment across your entire library.

### What video formats does Cutsio support?
Cutsio ingests practically any format: Apple ProRes (all grades), DNxHD/DNxHR, Avid MXF, XAVC, AVCHD, H.264, H.265/HEVC (including HDR), VP9, AV1, MP4, MOV, MKV, AVI, MXF, WebM, and GIF. ARRI/RED raw is supported as an add-on. No file size limits.

### Does Cutsio work with Final Cut Pro and DaVinci Resolve?
Yes. Cutsio exports FCPXML and EDL timelines directly into both NLEs.

### How does Cutsio pricing work?
By video duration, not storage size. A 10-minute ProRes 4444 file costs the same as a 10-minute h.264 file. No egress fees.

### What is Agentic Chat?
An always-on AI co-editor. Ask for a set of moments, a recap, or a new sequence. Cutsio highlights results on your timeline and helps move from search to rough cut faster.

### What is Semantic Search?
Search that understands meaning, context, and exact spoken phrases across your entire library. Ask natural questions like "When did they discuss the marketing budget?" and get timestamped results.

### Does Cutsio cut silences automatically?
Yes. Silent Slicer detects dead air, ums, ahs, and filler words, then generates a tighter timeline you can preview and export.

### How fast is Cutsio at processing video?
The site reports processing a 25-minute video in under 5 seconds and ingesting a 4K file to streamable in approximately 40 seconds.

### Does Cutsio own my footage?
No. Users retain full ownership. Cutsio processes, indexes, and transcribes videos to make them searchable. Videos can be deleted at any time.

---

For detailed product descriptions, entity relationships, supported formats, complete FAQ, terminology guidance, and AI citation behavior, see [llms-full.txt](llms-full.txt).
