Oracle puts OpenAI inside ERP: Fusion and NetSuite workflows run from ChatGPT, plus a managed AI security-testing service and a $12,000-credit SMB bundle

The source. The primary material is Oracle's official release Oracle Reinvents Enterprise Computing for the AI Era with OpenAI, published at Oracle AI World in Las Vegas on October 27, 2026. OpenAI published its counterpart the same day as OpenAI expands partnership with Oracle to bring its apps to ChatGPT. Product names, capability descriptions and quotes below come from those two official items; the pricing, timelines and module lists they do not contain are not invented here.

The overview. The two companies say they are bringing OpenAI models, ChatGPT and Codex together with Oracle applications, data and cloud infrastructure so AI runs directly across the systems and processes that underpin the global economy. It lands as four things: a workspace for getting work done inside ChatGPT; a managed service that uses frontier models to find and fix vulnerabilities; a tool that lets non-coders build enterprise applications; and a packaged bundle to get small businesses started.

Oracle AI Workspace. It puts OpenAI frontier models alongside Oracle applications, enterprise data and cloud services in a single user experience. Through plugins in ChatGPT, employees can reach information, complete approved business workflows and take action inside Oracle systems. Oracle says it is designed to connect Oracle Fusion Applications, Oracle NetSuite, Oracle AI Database, OCI and other enterprise tools and data sources. The key is the second half: AI goes into the flow of enterprise work while the governance, approvals and controls enterprise operations require stay in place. The chat box is the front door, not a way around the approval queue.

Oracle AI Security Mesh. This is a managed service that uses OpenAI cyber models to help customers find and fix vulnerabilities across OCI, other clouds and on-premises environments before attackers exploit them. It combines OpenAI frontier cyber models with Oracle's testing tools, independent validation, testing infrastructure and operational safeguards managed by Oracle, so customers do not have to build and operate their own defensive and offensive agent systems. The authorization logic is deliberately restrained: customers define where testing is authorized and control its scope, while Oracle manages the safeguards to reduce the risk of outages or business disruption during testing. In effect it turns the legal question of who is allowed to run penetration testing into a product feature.

Oracle App Composer. It lets employees turn ideas into enterprise applications that address recurring business problems, which Oracle says requires no coding expertise to build or maintain. Users build with Codex while Oracle manages deployment and operations on OCI. Applications connect to approved Oracle data and workflows and can use OCI services including OCI Kubernetes Engine and Oracle Autonomous AI Database. On top of that sit enterprise identity, permissions, networking, audit and application life cycle controls, so AI-generated applications can move from idea to secure enterprise deployment at scale. One telling detail: it supports APEXlang, a new Oracle technology pitched as keeping the speed of generative development without producing opaque application code that is hard to maintain.

The Oracle AI SMB Bundle. This is a packaged offering for small businesses, free to start, providing $12,000 of Oracle cloud credits each year. It bundles ChatGPT Business for daily work, Oracle AI Workspace for reaching business applications and data through ChatGPT, Codex and Oracle App Composer for building apps, and OCI for deploying them, plus NetSuite QuickStart as a foundation for core operations including invoicing and cash flow. Oracle notes that terms, conditions and restrictions apply. For large enterprises this line does not matter, but for the question of where a first AI workflow should start it provides a concrete price anchor: no entry cost, credits granted annually.

What the two sides said. Oracle chief executive Clay Magouyrk framed it as OpenAI transforming how enterprises solve business problems, from chat to coding, analytics and autonomous agents, with the new suite uniting those capabilities with Oracle applications, data and cloud infrastructure that businesses rely on to run critical operations. OpenAI president and co-founder Greg Brockman said more businesses are choosing OpenAI as it becomes easier to put AI to work through the tools and systems they already use, and that the expanded partnership brings frontier AI into everyday operations so teams can build software faster, strengthen security and take on more complex knowledge work. Both statements are notably restrained; neither uses the word replace.

Why the four ship together. None of the pieces is new on its own: chat boxes wired to ERP exist, AI vulnerability hunting exists, natural-language app generation exists. The structure only appears when you read them as a set. Workspace makes sure AI reads the right data. App Composer gives what AI writes somewhere to land. Security Mesh makes AI-written code get attacked by its own side before it ships. The SMB bundle lowers the cost of the first attempt. All four share one premise: AI output has to pass through the enterprise's existing identity, permission and audit systems, or none of them survives a compliance review.

One line that reads like a slogan deserves a closer look: the two companies say they share an ambition to set the standard for the agent-ready enterprise, partnering across the full innovation cycle to transform Oracle applications for agents and bring them into a unified experience in ChatGPT. The practical meaning is that Oracle is not merely exposing an interface so AI can read data, but reshaping its own applications into something agents can call directly. That is far more engineering than shipping a plugin, and it explains why the sentence sits in the release rather than in the product notes.

What is not said. Oracle published no pricing for any of the four, no general availability dates, no list of which Fusion and NetSuite modules are supported first, no detail on the authorization boundaries inside Security Mesh, and no specifics on how the $12,000 credits apply beyond a note that terms and restrictions apply. Security Mesh is described as using OpenAI frontier cyber models without naming the model, the testing methodology or a false positive rate. APEXlang is a new technology with no third-party maintainability data yet. These are exactly the items that carry weight in a procurement decision, and waiting for them before evaluating is reasonable.