How Oracle Is Shaping the Next Era of Enterprise AI: Key Themes from AI World 2026.

Oracle AI World 2026 took place in October and was extremely well attended. There was a palpable sense of anticipation among attendees. By now, most people have had their own experiences with AI, and most companies have dabbled with it. This year, they arrived ready for what’s next – and ready to get real about deploying AI.

This year, they could have called it “Agent World” given the emphasis on agents throughout the conference. Oracle has laid the groundwork for these AI Agents to do real work across both Fusion and non-Fusion platforms such as AI Data Platform (AIDP), Fusion Data Intelligence (FDI), and Oracle Analytics Cloud (OAC).

Below are the six main themes that struck me after attending the conference, followed by details on some of the key announcements.

Six Key Themes from AI World 2026

1. Turning Point for Agentic AI – we have been talking about AI Agents for a while, but I think we will look back in a year or two and cite this AI World as the point when AI Agents got real and adoption started to take off. One thing was clear at this AI World: Oracle has put all the puzzle pieces in place for Agentic AI, including Agentic Applications, to be developed, deployed and used safely and effectively. And this is true both within Fusion and outside of Fusion.
2. The Interface Is Changing – no matter who I was talking to about AI, whether they were more interested in executing business processes or getting analytical questions answered, the preference to interact through an AI agent came through loud and clear (including using Claude or ChatGPT as the interface). One Oracle customer I spoke to said he “did not want his accountants in Fusion.” At first I was puzzled, but then he clarified: he wanted his accountants in Claude, asking teams of agents to execute defined business processes and simply being informed of the results.
3. Enterprise Business Context Is King – large language models (LLMs) and the AI Agents that use them know nothing about anyone’s organization or business. Without that context, AI Agents are largely worthless (unless you want to know the weather or sports scores). This reality has driven the AI industry to find a solution to what is referred to as Enterprise Business Context. Oracle addressed this topic head-on and in a very significant way – establishing AI Data Platform as the Enterprise Context Layer driving Agentic AI development and adoption at scale.
4. Oracle AI Database and AI Data Platform Become Solidified and Entrenched as Multicloud – Oracle doubled down on the message that Oracle AI Database can run anywhere with Oracle’s full support. Additionally, Oracle took steps to make AI Data Platform multicloud, in the sense that it can work with metadata and data from other cloud data platforms like Snowflake and Databricks.
5. AI Across the Full Data Estate: Fusion, Non-Fusion, and Other Cloud Data Platforms – Oracle has enhanced FDI to more thoroughly embrace non-Fusion data sources by allowing a second semantic model in FDI. Oracle has more tightly integrated FDI with AI Data Platform. Oracle has enabled AI Data Platform to integrate and operate with other cloud data platforms. And Oracle has developed Deep Insight Agents that can be deployed in FDI, OAC, or AI Data Platform and can operate on any data, whether it is in Fusion, outside Fusion in OCI, or in another cloud data platform. As a result, you can draw a straight line from Fusion to FDI, to non-Fusion data sources in OCI, to data sources outside OCI, create Enterprise Business Context across all of that data, and have AI Agents use that context for Agentic AI execution across the full data estate – Fusion, non-Fusion OCI, and outside OCI in other Cloud data platforms.
6. Massive AI Unlock for Oracle E-Business Suite (EBS), PeopleSoft, and JD Edwards – Many Oracle customers are still running on EBS, PeopleSoft, or JD Edwards, and they have been concerned about being locked out of AI progress. The developments in AIDP, and the ability to create AI Agents and execute them against those systems, have dramatically changed the landscape for them. They can now use the Business Ontology, the Semantic Reasoning Engine, and the agent-building capabilities in AIDP to build agents that perform analytics or execute business processes – in a controlled and safe manner with full transparency. They can also create Deep Insight Agents against those systems.

Key Announcements from AI World

1. Fusion Claw – One of the biggest topics at AI World was Fusion Claw, which Oracle announced in late September, which is an agentic execution runtime that enables AI Agents and Agentic Applications in Fusion to conduct more complex, multi-step problem-solving work. Please see this article from my colleague Gary Marines on Fusion Claw capabilities: What is Oracle Fusion Claw?
2. Fusion Companion – ChatGPT as the Front End for Fusion Agentic Applications: With the introduction of Fusion Companion, Oracle is enabling users to interact with Fusion AI Agents and Agentic Applications through external conversational interfaces such as ChatGPT, while maintaining Fusion’s security, business context, and governed execution.Oracle Introduces New AI-Powered Ways to Execute Work Across Oracle Fusion Applications 
3. Fusion Data Intelligence (FDI) 
   • AI Data Platform Integration – AI Data Platform can now be provisioned right from within FDI. AI Agents built in AI Data Platform can act against data in FDI without moving the data out of FDI. This leverages the investment in FDI while enhancing FDI with the AI development and governance capabilities of AIDP.
   • Operational Analytics – the Oracle Transactional Business Intelligence (OTBI) subject areas from Fusion are now available from within FDI, so end users can view real-time Fusion analytics right alongside key performance indicators (KPIs), metrics, and historical data from FDI.
   • E-Business Suite Data Intelligence – Oracle announced a new module for FDI called EBS Data Intelligence that uses the FDI infrastructure and operates like FDI, with a prebuilt data pipeline, prebuilt data model, prebuilt semantic model, and prebuilt reports and dashboards – but with EBS as the data source.
   • FDI now supports a second semantic model – For those who do not know, a semantic model is where you define the relationships among database tables, calculations, metrics, hierarchies, and the subject areas that end users see in their reporting tool. A second semantic model means FDI can now include a semantic model defined over data that is not in FDI but in other databases or data lakehouses – either in OCI or in another cloud. Reports and dashboards can then be built that combine data from FDI’s data warehouse, Oracle Autonomous AI Lakehouse (ALH), with those other databases and data warehouses. This enables FDI to become a central console for analytics across all data sources – Fusion, non-Fusion and in other Clouds.
   • Tighter Integration Between Fusion AI Agents and FDI – FDI now includes a number of out-of-the-box agents and agentic applications that were built in Fusion AI Agent Studio but use metrics, KPIs, snapshots, and historical data that exist only in FDI.
   • Deep Insight Agent – see the OAC section below. This new capability in OAC works against the standard out-of-the-box FDI subject areas as well as any custom subject areas in FDI – including subject areas that are part of a second semantic model in FDI pointing to non-FDI databases and data sources.
   • FDI MCP Server: This capability existed before AI World, but it’s worth mentioning. FDI’s MCP server allows ChatGPT, Claude, and other MCP-compatible clients to serve as the front end for asking questions in natural language against FDI data. The answers leverage FDI’s semantic model, KPIs, and security.

Oracle Expands the AI Capabilities of Fusion Data Intelligence

4. Oracle Analytics Cloud (OAC)
   • Deep Insight Agent – Oracle announced a new AI Agent capability in OAC and FDI that answers complex questions with the answers including visuals and narrative summaries. The Deep Insight Agent is easy to set up and configure, and it is driven by a deep understanding of your Business Context. Oracle provides multiple pathways for giving the Deep Insight Agent this extensive Business Context.
   • Oracle Analytics Cloud MCP Server: This was available before AI World, but also worth mentioning. OAC’s MCP server allows Claude and other MCP-compatible AI clients to serve as the front end for asking questions in natural language against data in OAC, leveraging its semantic models and security.

5. AI Data Platform

In the world of AI, one of the biggest discussions right now – and one of the biggest problems to solve – is Enterprise Business Context and how to make LLMs and Agents aware of it.

As mentioned previously, LLMs and AI Agents know nothing about your organization or business. Unless you give the LLMs and agents Enterprise Business Context, it is like hiring an extremely smart intern who knows nothing about your business and therefore can do nothing.

To avoid confusion, it is important to define what we mean by Enterprise Business Context. The short answer is ‘everything’ – structured enterprise resource planning (ERP), supply chain management (SCM), and customer relationship management (CRM) data, policy documents, supplier and customer contracts, regulatory documents, Slack channel discussions, emails, etc.

Oracle has defined AI Data Platform as the point where all an organization’s Enterprise Business Context comes together to be leveraged by AI Agents.

There were several announcements at AI World related to AI Data Platform and Enterprise Business Context which are summarized below:

• Business Ontology – this enables you to define the semantics of how your business is organized and operates. Oracle will generate version 1 for you, based in part on metadata consumed from your systems and databases, and you can then modify the Business Ontology as needed. This is equivalent to creating an instruction manual for your business for the LLMs and the agents that use them.
• Uploaded Documents as Part of the Enterprise Business Context – enterprise documents can be uploaded into AIDP and incorporated into the Enterprise Business Context, giving LLMs and AI Agents relevant business knowledge to support more accurate, context-aware reasoning and decision-making.
• Semantic Reasoning Engine – the Semantic Reasoning Engine identifies and assembles the relevant business context before passing it to the LLM, allowing the LLM to focus on analyzing the information and answering the question.
• AI Data Platform Is Multicloud and Consumes Metadata from Other Cloud Data Platforms – AIDP can now consume the metadata of other cloud data platforms and build a Business Ontology on top of that metadata. This allows AIDP to be used as the platform to build and execute AI Agents against other cloud data platforms like Snowflake and Databricks.
• Deep Insight Agent Acting on Data in Other Cloud Data Platforms – A Deep Insight Agent built in AIDP can leverage business context from other cloud data platforms such as Snowflake and Databricks to analyze data, investigate business questions, and recommend actions.
• AI Agent Hub – Oracle is providing a ChatGPT-like conversational interface within AI Data Platform, allowing business users to interact with Oracle and third-party AI agents from one place. Customers can still use ChatGPT or Claude as their AI interface if they prefer, but Agent Hub provides an alternative within their OCI environment, using their own security and access controls.
• AI Registry – Enterprise-Wide AI Agent and Asset Discovery – AIDP includes a centralized AI Registry that catalogs AI agents, MCP servers, tools, and reusable skills across the enterprise, making them discoverable and accessible through Agent Hub and other AI interfaces. At this time, Oracle Fusion AI Agents, whether out-of-the-box or custom-built, are not automatically added to the AIDP AI Registry, but they can be registered manually.
• AI Governance – Enterprise-Wide Agent Monitoring and Oversight (Coming Soon) – Oracle announced a new AI Governance Framework that will enable organizations to monitor AI agent performance, quality, usage, costs, and drift across the enterprise. It will also support agent testing, certification, and compliance with organizational policies, helping enterprises manage their growing ecosystem of AI agents.

Oracle AI Data Platform Brings AI-Assisted Business Context to Enterprise AI

Oracle AI Data Platfom: Enterprise Context for Agentic AI at Scale (TK Anand’s Keynote Speech)

Final Thoughts – My Suggestions for Taking Action

If I had to sum up Oracle AI World 2026 in one sentence, it would be this: the pieces are now in place for AI Agents to do real work across the entire enterprise. Oracle has connected Fusion, FDI, OAC, AI Data Platform, and other cloud data platforms, and it has put Enterprise Business Context at the center of it all.

For most organizations, the question is no longer whether AI Agents can deliver value, but where to start and how to make it happen. I would start by identifying and prioritizing Agentic AI use cases, but at the same time make sure the organization has the infrastructure to safely develop, deploy, and monitor AI Agents and teams of agents. Security and guardrails need to come first.

I would also put in place an automated mechanism to continuously capture Enterprise Business Context and make it available to LLMs and agents. Equally important is having full visibility into what agents are doing, with the transparency and governance needed to manage them across the enterprise.

Not all Agentic AI use cases are the same. I would distinguish between Analytics AI, where agents help answer questions and provide insights about the business, and Business Process AI, where agents actually take action and update operational systems. I would also distinguish between use cases that require a single agent and those that benefit from a team of agents working together.

Finally, I would be realistic about the organization’s AI maturity. Start with simpler use cases, build experience and confidence, and gradually move toward more complex, multi-agent applications. The goal isn’t just to deploy AI Agents. It’s to build the capability to deploy them safely and at scale, delivering real business value.

And for those still running EBS, PeopleSoft, or JD Edwards, the message from this AI World is clear: you are no longer locked out of AI.

Thank you to everyone who stopped by the Apps Associates booth, and to our Apps Associates clients and Oracle colleagues who shared their perspectives throughout the week. If you would like to talk through any of these themes or what they could mean for your organization, feel free to reach out to me at [email protected] or via LinkedIn messaging.

About the Author

Myles is the Vice President of Data, Analytics and AI at Apps Associates. He has 30 years of experience designing, building and deploying large scale data warehousing and business intelligence systems. Myles is a thought leader in the area of data strategy, modern data architecture and advanced analytics and frequently speaks at industry conferences. Before getting back into consulting about 9 years ago, he held senior executive positions in the Financial Services and Medical Device industries. Myles has significant international experience having served in Director roles in Tokyo, London and Hong Kong. He specializes in helping companies extract value from their data and transform themselves into data-driven organizations.