Cloudinary has introduced an Agent Experience layer designed to let AI agents build, manage, and optimize visual media workflows with less human configuration. The platform combines media infrastructure, APIs, governance, and AI-agent tooling so developers can delegate implementation and operational tasks while retaining control over application design and policies.

AI Agents Can Provision Their Own Environment

Cloudinary’s new “Claimable Cloud” allows AI agents to provision a fully featured environment using Cloudinary’s production infrastructure without requiring traditional manual account setup.

Agents can use the environment to:

• Evaluate Cloudinary capabilities.

• Configure visual media workflows.

• Integrate media functionality into applications.

• Test implementations.

• Begin delivering functionality autonomously.

Developers can later claim the environment and continue development without losing the work completed by the agent.

Making Visual Media Understandable to AI Agents

Traditional media systems often expose images and videos primarily as files or URLs. Cloudinary is adding structured context around each asset so agents can understand how media should be used.

Each visual asset can include information such as:

• Searchable metadata

• Governance rules

• Permissions

• Usage rights

• Relationships to other assets

• Delivery information

This allows agents to search for appropriate content, determine whether it can be used, and execute media workflows according to predefined policies.

Automating Visual Media Workflows

Cloudinary describes several potential agent-driven workflows.

For a marketplace, an agent could evaluate user-generated images against brand and compliance requirements, send non-compliant assets for human review, and publish approved content automatically.

For social media automation, an agent could retrieve approved campaign assets, generate platform-specific formats and aspect ratios, apply brand-approved overlays, and schedule content across channels such as Instagram, TikTok, and LinkedIn.

AI-Generated Content Stays Within the Platform

Cloudinary’s Image Generation and Image to Video capabilities can create content directly within the platform. Generated assets are automatically classified, tagged, and stored according to existing policies.

This means AI-generated images and videos can receive the same metadata, governance, permissions, and workflow treatment as existing media.

Infrastructure for Agent-Native Development

Cloudinary supports an agent-oriented development model through MCP, agent-native tooling, LLM-readable documentation, Skills packages, AI-optimized SDKs, and deterministic APIs.

These capabilities provide agents with access to core visual media functions including:

• Storage

• Search

• Transformation

• Optimization

• Governance

• Delivery

The underlying model separates responsibilities between developers and AI agents: developers establish application requirements and policies, while agents can execute the technical media workflows within those boundaries.

From Media Management to Agent Execution

Cloudinary’s launch reflects a shift from visual media platforms designed primarily for human desired outcomes while agents handle more of the implementation, media processing, governance, and delivery work required to achieve-operated workflows toward infrastructure that AI agents can directly understand and operate.

The goal is to allow developers to describe desired outcomes while agents handle more of the implementation, media processing, governance, and delivery work required to achieve them.