Playcut

1 review

Playcut is an agent-native AI content studio for marketing teams, agencies, and DTC brands. It ships a full Model Context Protocol (MCP) server — many tools covering actor generation, image, video, voice cloning, brand kits, and batch workflows — so the entire studio is drivable from Claude, Claude Code, Cursor, Windsurf, Zed, and any other MCP-compatible AI client. A REST API covers the same ground for ChatGPT custom GPTs, n8n, Zapier, Make, and custom backends. Teams brief an AI agent in plain language and campaign assets come back, rather than clicking through a studio UI shot by shot. The core of the product is the custom AI actor. Teams build actors unique to their brand — not pulled from a stock library every competitor is also using — and hold the same face, body, wardrobe, and styling across every still, video, UGC clip, and product shot, so one actor carries an entire campaign instead of breaking after the third image. Customers retain ownership of the actors they build and of every generation they create, with commercial use included on all plans and no watermark on any tier. Customer content is never used to train foundation models. Behind both surfaces sit 13 model surfaces from five vendors — Google, xAI, ByteDance, Alibaba, and HappyHorse — spanning image, video, music, and voice. There is no automatic model router: teams pick the model on every generation or take the capability default, so output is never swapped behind their back. Multi-brand kits hold colors, typography, logos, and voice guidelines. Shared team folders and per-user private folders keep in-house and multi-client agency work separated, with creator-only edit rules on shared work.