Generative media for agents
Also known as: agent-callable creative API, creative API for AI agents, MCP for media, media rendering infrastructure for agents
Generative media for agents is the category of image, video, audio, and 3D APIs designed to be called by autonomous AI agents — not humans clicking a UI. They expose tools through MCP, return structured outputs an LLM can parse, and tolerate retries, long jobs, and failure modes only an agent encounters.
Example
An agent running in Claude Desktop receives "generate a hero banner for SS26 sage linen, 16:9, four variants". It calls lamina.create with a brand id and an idempotency key, streams progress, picks the highest-scoring asset, and posts it to the Shopify product page — without a human in the loop.
Why the category exists
Most generative-media APIs were designed for humans with screens. An agent doesn't have a screen. It has a tool list, a budget, a step counter, and a non-zero chance of timing out and retrying. A media API built for agents looks different from one built for designers: it accepts idempotency keys, streams phase events, returns structured asset metadata, and exposes itself through Model Context Protocol so any MCP-aware client — Claude, Cursor, Windsurf, VS Code — can invoke it natively.
How it differs from a generative-media API
A generative-media API exposes a model: pass a prompt, get an image. A generative-media API for agents exposes a creative app — a higher-order tool that routes across models, attaches brand context, evaluates the result, and delivers it where it needs to go. The agent never picks the model; it picks the outcome. That keeps prompt logic out of agent code and survives the next model upgrade.
What a typical call looks like
An agent in Cursor receives the task "generate a hero image for the SS26 launch". It calls the lamina.create MCP tool with a brief, a brand id, and an idempotency key. The server returns a job id and an ETA. The agent calls lamina.track to stream phase events, lamina.evaluate to score the assets against the brand kit, and lamina.distribute to push the keepers into Shopify and Sanity. The whole loop is four tool calls — no UI in the path.
References
Related terms
Creative API
A Creative API is a developer interface that wraps multiple generative models, brand context, evaluation, and distribution behind a single endpoint. Instead of calling raw model endpoints — Sora, Veo, FLUX, Runway — directly, you call a creative app like productShoot or virtualTryOn and the API handles routing, evaluating, and delivering the result.
MCP server
An MCP server is a process that exposes tools, resources, and prompts to an LLM client over the Model Context Protocol — Anthropic's open standard, donated to the Linux Foundation in December 2025. MCP servers let agents in Claude, Cursor, Windsurf, and VS Code call your API as native tools.
Agent-callable API
An agent-callable API is one designed to be invoked by an autonomous LLM agent rather than a human developer. It exposes typed tool schemas, returns structured outputs an agent can parse, supports idempotency keys for safe retries, and handles long-running jobs through streaming or polling instead of blocking.
Creative App
A creative app is a reusable, named generative-media workflow — productShoot, virtualTryOn, brandFilm — that bundles model routing, prompt scaffolding, brand context, and output formatting into a single callable unit. You call apps, not models, so the same call survives a model upgrade and produces consistent on-brand outputs.