Oracle AI World ran from 25 to 28 October in Las Vegas, and the news that matters most to application customers came on the second day. On 26 October Oracle announced an agentic AI platform to speed up the implementation and modernisation of Oracle Fusion Cloud Applications, alongside new ways for AI to carry out work inside Fusion.
Conference announcements are written to impress. This article sets out what Oracle announced, what is confirmed and what is still a roadmap, and what a buyer should ask before any of it appears in a renewal quote.
AI Success Navigator and faster implementations
The headline product is Oracle AI Success Navigator. Oracle says it combines specialised AI agents, implementation best practice, expert guidance and training. The first agents help customers analyse their current environment (a Discovery Agent) and turn business processes into application designs for review (a Design Agent), and Oracle says more are planned.
Oracle quoted an early adopter, Lexington Health, saying that the agentic approach cut certain design and configuration effort by 80 percent. That is one customer’s claim about specific tasks. It is not a promise about total project cost, and it should be treated as an upper marker, not an average.
Oracle also described a free resource for implementation work. The Fusion Setup and Configuration Registry is publicly available and gives teams and AI tools a common foundation for configuration, which suggests Oracle wants configuration knowledge to be reusable and machine-readable.
Agents inside Fusion
Day-to-day work is changing too. One recap says Oracle showed the 25 Fusion Agentic Applications it launched on 29 September, taking its agentic portfolio to 75, and introduced Fusion Companion and Fusion Agentic Home. Companion is described as a layer that coordinates transactions, workflows, approvals and agents, and Agentic Home as the place where users delegate work and track it.
Oracle also announced Oracle AI Workspace, designed to let employees complete approved workflows across Fusion, NetSuite, Oracle AI Database and OCI from within ChatGPT. That brings a second supplier into the picture, which matters for data handling and for who you contract with.
What is confirmed and what is not
Not everything announced is available. One recap describes Oracle’s Deep Trust AI Architecture as a blueprint and says several components remain on the roadmap. Another reports that Fusion Companion and Agentic Home are part of release 26D. Before you plan around any feature, ask Oracle whether it is generally available today, in preview or still planned, and in which release.
The same applies to the multicloud and database news. Oracle highlighted Oracle AI Database running inside AWS, Azure and Google Cloud data centres, and Snowflake arriving on OCI as a native service. These are useful options, and each comes with its own commercial terms that the announcement does not describe.
Five questions to ask Oracle
First, how are AI agents charged? One recap says Fusion uses a separate token-based meter called AI Units. Treat that as something to confirm in writing, with the rate, the unit and any expiry. Second, which of the announced features are included in the subscription you already hold, and which are new purchases?
Third, what changes for your implementation partner? If Oracle’s tools reduce effort, your statement of work should reflect it, and you should know who pays for any Oracle tooling the partner uses. Fourth, if a third-party AI service such as ChatGPT connects to your Oracle data, which agreement governs the data, and where is it processed?
Fifth, what is the timeline? Oracle’s fiscal second quarter ends on 30 November, so offers linked to these announcements may arrive with a deadline. Ask for the terms in writing and compare them with your current agreement before accepting a date that suits Oracle’s calendar.
If you run E-Business Suite
The same recap reports that E-Business Suite 12.2 is certified on Oracle AI Database 26ai with conditions, and that EBS Data Intelligence is generally available for analytics across EBS and Fusion during a phased move. If that describes you, check the conditions for your platform and version before you assume the certification applies.
What it means for implementation partners
If Oracle’s tools do reduce effort on design and configuration, the commercial effect lands first on implementation contracts. A fixed-price statement of work written before these tools existed may be priced on assumptions that no longer hold. A time-and-materials contract will pay for hours that could now be shorter.
Ask your partner which Oracle tools they plan to use, what difference they expect them to make and how any saving will show in the price. Ask also who pays if the tooling is metered or sold separately. A partner who cannot answer these questions yet is not necessarily being evasive, since the tools are new, but you should agree how the question will be settled before the project is under way.
Separating announcements from entitlements
A useful habit is to keep two lists. The first is what Oracle has announced. The second is what your contract entitles you to today. Many disputes start when a customer believes something is included because it appeared in a keynote, while the order form says otherwise.
When your account team refers to a new capability, ask which order form line it falls under. If the answer is not clear, assume it is not included until you have it in writing. Update the second list whenever your agreement changes, so that you always know what you own.
Conclusion
Oracle AI World showed a clear direction: agents that help implement Fusion, agents that do work inside it and a closer link to OpenAI. Some of this is available now and some is still a plan.
Ask which parts are generally available, how they are priced, what changes in your partner contracts and who handles your data. Get the answers in writing before they reach your renewal.
