Summary

• Oracle AI Workspace is being developed to connect ChatGPT with enterprise applications, data and approved business workflows.
• The partnership also introduces plans for Oracle App Composer and AI Security Mesh, using OpenAI technology for application development and security testing.
• AI Workspace remains forthcoming, while Oracle Marketplace purchasing of OpenAI services and the new SMB programme have US-specific eligibility restrictions.

Oracle and OpenAI are developing an enterprise workspace that will connect ChatGPT with business applications and operational data, allowing employees to retrieve information and coordinate approved activities across systems normally accessed through separate interfaces.

The proposed Oracle AI Workspace will bring together Oracle applications, enterprise information and OpenAI models through a common interface. Its intended capabilities include accessing relevant business records, identifying outstanding work and coordinating actions subject to an organisation’s existing permissions and approval requirements.

The companies have also introduced plans for application development and security services using OpenAI technology. Oracle App Composer is intended to help employees build and share applications, while Oracle AI Security Mesh will use specialised cybersecurity models to identify weaknesses in authorised systems.

Oracle AI Workspace remains under development, with no confirmed general availability date. Although the partnership includes products and commercial arrangements already in use, the intended workspace workflows should not be presented as capabilities currently deployed across Oracle’s customer base.

Connecting information across enterprise applications

Large organisations use different applications to manage finance, procurement, human resources, planning and operations. Those systems may exchange information, but employees frequently need to navigate separate interfaces or request reports to understand how activities across departments relate to one another.

Oracle’s proposed workspace aims to provide a conversational route into those systems, allowing an authorised user to describe a task rather than identify every application and report needed to complete it.

A finance employee examining month-end accounts, for example, might need to inspect outstanding purchase orders, identify discrepancies and establish which approvals remain incomplete. The intended interface would bring relevant information together while preserving the permissions associated with each underlying application.

The approach involves more than returning information in response to a question. Oracle wants AI agents to coordinate approved workflows, potentially initiating actions within connected business systems when the relevant conditions have been satisfied.

That creates a distinction between reading a record and changing it. An employee permitted to view a supplier invoice may not be authorised to approve payment or modify bank details, even when the activities form part of the same broader process.

The companies therefore emphasise unified sign-in, administration and the retention of organisational access controls. A conversational interface cannot safely bypass the approval procedures governing the applications from which it draws information.

Reliable implementation will also depend on maintaining the context behind business records. An outstanding payment, for example, may be associated with purchasing rules, contractual conditions or information held by another department that an AI system must account for before recommending an action.

Application development becomes part of the offering

Oracle App Composer extends the partnership into software creation. The proposed service uses Codex and related AI tools to help employees develop applications and connect them to approved enterprise information and processes.

Oracle intends to manage deployment and operation through Oracle Cloud Infrastructure, while IT teams retain control over permissions, connections and other administrative requirements.

This approach targets a common problem in business software. Employees can increasingly create useful applications without specialist programming skills, but sharing those tools across a department introduces requirements for access management, maintenance, security testing and ongoing support.

An application designed to track invoice exceptions may initially be useful to one employee. Once colleagues depend on it, the organisation needs to know which information it accesses, who is responsible for changes and whether it can continue operating reliably.

Oracle describes App Composer as a way to move those applications into a managed environment rather than allowing them to develop into an uncontrolled collection of individual tools.

The company has also outlined APEXlang, a technology intended to support the development of applications that remain understandable and maintainable after AI-assisted creation.

These offerings represent Oracle’s product direction, and their release status must be assessed individually. Announcing a managed development platform does not establish that every described function is available to all customers.

Security testing becomes another AI application

Oracle AI Security Mesh will use specialised OpenAI cybersecurity models to discover, investigate and help prioritise vulnerabilities in applications and infrastructure.

The planned service combines automated analysis with testing intended to establish whether a weakness presents a meaningful risk. Oracle describes a managed environment in which customers authorise the systems being examined and retain control over consequential remediation decisions.

Automated security assessment can help teams examine complex applications and identify relationships between weaknesses that might otherwise require substantial manual investigation.

However, security testing can also affect live systems if intrusive techniques are used without suitable controls. Organisations must therefore define the permitted scope, operational restrictions and procedures for reviewing proposed corrective actions.

Oracle’s design places those activities within a customer-authorised process rather than offering unrestricted automated access to infrastructure. The intended approach will require validation through deployment, particularly where systems support essential operations or contain sensitive information.

Oracle reports substantial internal use

The expanded collaboration builds on an existing commercial relationship involving the OpenAI API, ChatGPT Work and Codex.

Oracle reports approximately 130,000 active ChatGPT users and more than 95,000 active Codex users within its organisation. The companies also cite an internal recruitment research process in which preparation that previously required two to four days can now take approximately 15 to 20 minutes using ChatGPT Work.

Those figures describe reported internal adoption and a particular workflow. They do not establish equivalent productivity gains across other enterprises or prove the effectiveness of the proposed Oracle AI Workspace.

Existing use nevertheless provides experience with AI adoption inside a large software business, including the management of accounts, permissions, costs and the information required to support everyday tasks.

The companies are now extending their relationship from the supply and use of AI tools towards deeper integration with applications that hold operational business data.

European access remains uncertain

Oracle customers in the United States can already purchase eligible OpenAI services through Oracle Marketplace, using established procurement arrangements and qualifying Oracle Universal Credits.

That purchasing route currently applies to US customers and should not be treated as a general commercial offering available throughout Europe.

Oracle has separately introduced an AI SMB Bundle intended for qualifying smaller businesses. The programme combines access to AI tools and Oracle services with an allocation of up to $12,000 in OCI credits annually, subject to eligibility and renewal conditions.

The programme is restricted to qualifying privately held US businesses, and its component services have separate access conditions. Inclusion in the bundle does not mean that every proposed product is immediately available or that use is unrestricted.

For European enterprises, future availability of the broader integration programme will depend on regional product launches, contractual terms and the technical arrangements governing connected applications.

Organisations will also need to assess where information is processed, which service providers can access it and how identity and approval controls operate across the proposed workspace.

Oracle AI Workspace is intended to reduce the separation between business software and conversational AI, but the company has not committed to a release date or final specification. Its commercial value will become clearer when customers can test the integration against existing applications, access restrictions and operational workflows.