Loova brings product shoots and video ad workflows into AI agents through MCP

Loova is moving visual production into the conversation where an AI agent is already planning the work. Its November 3 MCP announcement describes image generation, video generation and specialized creative workflows that compatible agents can call, turning a product reference into assets such as a studio photograph, listing images or a promotional video.

The launch has two layers. One exposes generation models through Loova; the other organizes several production steps around a particular outcome. The company names models including Seedance, Kling and Wan, but Loova is the service and workflow integration around those models, rather than a new foundation model of its own.

From a product image to a campaign asset

The announcement describes nine initial specialized workflows across product imagery, video and advertising, image adaptation and character-focused production. Its English live page illustrates Product Photoshoot, Product Listing Images, UGC Ad Creator, Ad Resizer and Product Video Maker with sample requests.

That gives the integration a more concrete pitch than simply adding another image tool to an agent. A marketer could reuse one product reference for a lifestyle photograph, a coordinated listing set and a showcase video, while keeping the brief inside the agent’s conversation. The company presents this as a way to reduce switching between production applications; AITrending has not measured time savings or reproduced the examples.

The “Viral Video Clone” workflow is particularly explicit about review. Loova says users supply a reference video and replacement product or avatar imagery, then review and edit an adaptation script before rendering. Its English MCP page says that checkpoint comes before spending credits on generation. The name does not establish that an output will go viral or reproduce a reference exactly.

Connecting still requires an account and authorization

The live English setup page presents separate client routes for ChatGPT and Codex, Claude, Claude Code and a CLI. Its initial route identifies the remote MCP endpoint and an official creative-skills package, then directs the user to sign in to Loova in a browser and return to the agent to verify the connection. Support depends on the client’s integration method; the page does not establish one minimum version that covers every host.

Generation uses Loova’s commercial service. Its pricing page offers credit-based plans and model-specific unlimited offers with differing limits, so connecting an agent should not be read as making every model or workflow free. A reference asset used in generation enters that service’s workflow, rather than remaining a purely local image-editing task.

Loova also outlines future animation, short-form drama, storyboard and video-assembly work. Those are expansion plans. The current announcement’s practical contribution is a connection between an agent’s planning conversation and reusable visual-production workflows, with a visible human review point for adapting a video.