Guide to publishing AI agents in Gemini Enterprise and Google Cloud Marketplace

Developers can now publish and commercialize third-party AI agents through Google Cloud Marketplace and deploy them to the Gemini Enterprise app. This guide outlines a step-by-step process for creating integrated, marketplace-ready solutions.

The A2A Agent Card is a JSON file used to declare an agent's capabilities, supported authentication and authorization methods, and service endpoints. The Gemini Enterprise app uses this card to display agent information, locate endpoints, discover entry points for messages, and determine authentication requirements.

The end-state architecture bridges Google Cloud Marketplace billing, identity provider (IdP) security, and the Gemini Enterprise Agent Platform. Key components include the customer project for discovery and interaction, the partner project hosting the agent and marketplace handler, and a partner Marketplace project managing procurement and event notifications.

To sell agents on Marketplace, organizations must join the Google Cloud Partner Network, review Agent-as-a-Service listing requirements and the Marketplace Vendor Agreement, and nominate their agent through their Google Cloud representative. Agents must also comply with agent-specific mandates.

Agents must adhere to the A2A protocol and potentially A2UI for rich user interfaces. A mandatory A2A Agent Card (JSON file) must declare capabilities, authentication methods, and service endpoints. Agents must support public access or OAuth 2.0 Authorization Code Grant Flow, and integrate with Procurement APIs and Pub/Sub for entitlement management.

The landscape of software is evolving towards Agents-as-a-Service (AaaS), where AI agents interoperate using standardized protocols like Agent2Agent (A2A) and are orchestrated through platforms such as Gemini Enterprise Agent Platform. The focus is on engineering autonomous agents capable of executing complex, multi-step workflows for tangible business value.