Enterprise Context as Primitive

Onix and Google Cloud just expanded a partnership built around a strange-sounding idea: before an artificial intelligence agent can do useful work, it needs a company’s internal map of what things mean. Onix says that map sits inside its Wingspan platform as a “Semantic Twin,” and the two companies expect the broader collaboration to drive more than $500 million in Google Cloud consumption. (onixnet.com) The immediate problem is simple: one team’s “customer” is often another team’s “account,” and a third team may split the same person across billing, loyalty, and support systems. When an artificial intelligence agent pulls from all three, it can follow the wrong rule even if the underlying data is technically correct. (cloud.google.com, docs.cloud.google.com) That is why companies talk about an ontology, which is just a formal list of what entities exist and how they relate to each other. In a retailer, that list can connect a brand to a region, a product to a policy, and a partner to the contract terms that limit what an agent is allowed to say or do. (thehindubusinessline.com, cio.economictimes.indiatimes.com) Onix’s pitch is that this context layer should be shared across the company instead of rebuilt inside every chatbot, search tool, and workflow bot. The company says its Semantic Twin gives agents “enterprise context and business ontology,” which means the same definitions travel with the agent instead of being re-created by each product team. (onixnet.com, prnewswire.com) This gets more important in multi-brand companies, where one platform may sell premium, discount, and wholesale products under different names. A return policy that is valid for Brand A in California can be wrong for Brand B in Germany, so an agent needs the brand and jurisdiction attached to the answer before it generates anything. (cio.economictimes.indiatimes.com, cloud.google.com) Google Cloud’s side of this story is the infrastructure for building, connecting, and governing agents. Google says Gemini Enterprise can connect agents to company data in systems such as Google Workspace, Microsoft 365, Salesforce, SAP, and BigQuery, while giving administrators centralized controls over deployment and oversight. (cloud.google.com, docs.cloud.google.com) So the partnership is not just about one new model. It is about putting Onix’s business-definition layer on top of Google Cloud’s data and agent stack, so a company can move from raw records to agents that know which brand, policy, and partner rule applies in a given situation. (thehindubusinessline.com, onixnet.com) Onix says that setup is already tied to “thousands of AI agents” running in production at Fortune 500 companies, though that figure comes from the company itself rather than an independent audit. The company also says Wingspan can move projects from concept to production up to three times faster than traditional consulting approaches. (onixnet.com, mobileworldlive.com) The bigger shift is that enterprise artificial intelligence is starting to look less like a model race and more like a context race. If every agent can already write a decent sentence, the harder advantage is knowing that “gold customer,” “reseller,” and “do not discount” mean the same thing across every system the company owns. (cloud.google.com, onixnet.com)