The Universe
of Commerce Agents

Commerce Universe is Divante's set of AI commerce agents that plug into the eCommerce platform you already run, through one shared integration layer called the Golden Connector. There are six agents, each built for a specific challenge: portAIge (platform migration), ProductLens (product content), Figurehead (imagery), Lodestar (merchandising), Tradewind (personalisation) and Sounding (assurance). Every agent grew out of work Divante first hand-built for real clients over 18+ years of eCommerce engineering.

Agentic commerce is a model of eCommerce in which AI agents act on behalf of people. On the shopper's side, they search for products, compare offers and complete purchases. On the merchant's side, they run tasks such as migrating data, writing product content or adjusting merchandising. For online stores, this means catalogue, prices and availability have to be readable and actionable not only for human shoppers but also for AI agents, through open protocols such as UCP, ACP and MCP.

A commerce agent is an AI system that performs a specific eCommerce task with a degree of autonomy, using data and tools from the store platform. Unlike a chatbot, which only answers questions, an agent takes actions: it maps data during a migration, generates product content or adjusts merchandising rules. In Divante's Commerce Universe, each agent is dedicated to one challenge and connects to the store through the Golden Connector.

portAIge is Divante's AI agent for eCommerce replatforming. It brings AI into the parts of a platform migration that usually consume the most time and carry the most risk: inventorying customisations and integrations on the legacy platform, mapping data between the old and the new platform's schemas, and checking feature parity before any traffic is moved. It is built on the same methodology Divante has used in 150+ production migrations.

The Golden Connector is Divante's integration layer between AI agents and a commerce platform. It reads each store's own data schema and maps it onto a common contract, so an agent is built once instead of once per platform. It natively supports the three main agentic commerce protocols, UCP, ACP and MCP, so your agents keep working regardless of which protocol becomes the market standard.

They are open protocols that define how AI agents communicate with commerce systems. UCP (Universal Commerce Protocol), co-developed by Google and Shopify, covers the full shopping journey from product discovery to checkout and order management. ACP (Agentic Commerce Protocol), developed by OpenAI and Stripe, focuses on agent-led checkout in the ChatGPT ecosystem. MCP (Model Context Protocol) is a general standard that gives AI models access to tools and data. Since it is not yet clear which will dominate, Divante's Golden Connector supports all three in one integration layer, so you do not have to bet on a single protocol.

No. Every agent connects to the platform you already run through the Golden Connector, so the store does not have to be rebuilt to accept an agent. A single integration layer also means one place to secure and monitor, instead of a separate point integration for every agent that can break on a platform upgrade.

Yes. Most teams start with the agent that solves their most pressing challenge, often portAIge when a replatforming is on the horizon. The Golden Connector is included as the foundation of every agent engagement, and it is also available on its own as an integration audit or build for teams assembling their own agents.

Start with one conversation with Divante's team. Tell us what is slowing your team down, and we will tell you honestly whether an agent is the right fix or whether something simpler would do.