Oracle spent twelve months building a coherent AI platform for Fusion — one studio, one button, and a growing line-up of agents and agentic applications across Planning, Procurement, Manufacturing, Inventory, Warehouse, PLM, and Logistics. This is what Oracle has actually shipped and where Oracle says it is going next — drawn from Oracle’s own readiness notes and announcements, not vendor slides.
By Creative Flakes Communications — an Oracle Fusion & E-Business Suite implementation and support practice.
Fusion AI Agent Studio. Announced March 20, 2025. Expanded with an Agentic Applications Builder in March 2026. Supports multi-agent workflow orchestration, contextual memory, content intelligence, multimodal processing, monitoring & observability, and ROI measurement. Every Oracle AI Agent described in this magazine runs on it.
Home with Ask Oracle is the home-and-global-navigation surface — a Fusion-wide pattern in the bottom-right of every Redwood screen, desktop and mobile. Inside products, agents launch from in-page Ask/ellipsis buttons (WMS), Ask AI from More Actions (SSP), or AI Assist on action menus (Planning). Predictable vocabulary. No bespoke chat widget per module.
Per Oracle’s May 2026 AI Features Report: 178 SCM AI items across Agent, Agentic App, Generative, and Predictive types — 54 named agents and 10 agentic applications in the 26A and 26B releases alone, plus the WMS predictive features that have been shipping since 22D. This issue covers the named agents in Warehouse, Planning, and Procurement, plus the Logistics Execution Command Center.
What follows is the honest, named list.
Oracle’s enterprise AI story is easy to dismiss as catch-up. Reading the announcements in order says something different: a deliberately sequenced platform build, from developer studio to data platform to agent marketplace to agentic applications — shipped publicly between March 2025 and April 2026.
Oracle is not shipping AI features. Oracle is shipping an AI operating model.
Fusion AI Agent Studio is Oracle’s complete development platform for building, connecting, deploying, and running AI agents across Fusion Applications. Announced March 20, 2025. Expanded March 24, 2026 with the Agentic Applications Builder — a no-code way to compose multi-agent workflows using natural language instead of traditional coding.
Every Oracle-shipped agent named in this magazine runs on it. Customers and partners build agents on the same platform.
Why the platform matters more than any single agent. The conversation with customers is no longer “does Fusion have an AI feature for X?” — it’s “can we build (or buy) an agent to do X, and will it work with the rest of Fusion?” Agent Studio is Oracle’s answer to that shift.
Oracle has made a deliberate UI choice. Rather than scattering AI widgets per module, Fusion exposes AI through a small set of consistent surfaces. The home-and-global-navigation surface is Home with Ask Oracle — a Fusion-wide pattern introduced in 25C and reaffirmed in the 26B Warehouse Management What’s New: “after the Redwood screen opens, click the Home with Ask Oracle button in the bottom right.”
Within products, agents launch from the surface that fits the workflow: WMS uses an in-page Ask/ellipsis pattern; Self-Service Procurement uses an “Ask AI” entry from More Actions; Planning uses AI Assist on action menus. The point isn’t one global button — it’s a small, predictable vocabulary of invocation patterns instead of a different chat widget per module.
Also appearing across Redwood
Across Redwood Desktop and Redwood Mobile, an ellipsis “Ask” button appears on screens that host AI Agents. It’s the second invocation pattern — the in-page equivalent of the home-page Ask Oracle button.
The hardest part of enterprise AI adoption is not the model. It is the discovery: knowing that an agent exists for a task and where to find it. One button — the same one, in the same place, on every screen, across every module — turns discovery into muscle memory.
Pause — before we count agents
One platform. A small set of invocation patterns. Many, many agents.
The rest of this magazine is a tour. Module by module, Oracle has named agents and shipped them into Fusion SCM between the 26A release in January 2026 and the agentic-applications announcements in April 2026. What follows is the honest, named list.
The three warehouse agents are where Oracle’s Fusion SCM AI story first became concrete. All three are documented in the 26A readiness notes, all three are conversational, and all three launch from the “Ask Oracle” button in the Redwood UI.
Embedded in the Wave Inquiry screen. Summarises a wave run in natural language, flags urgent issues — shortages, resource gaps, allocation failures — and lets a planner ask follow-up questions without leaving the screen.
Readiness note f42424
Identifies expired and near-expired lot inventory across reserve, active, and outbound locations. Can lock containers to prevent further allocation once flagged. Critical for food, pharma, chemicals, and anywhere lot shelf-life drives cost.
Readiness note f42576
Chat over WMS tasks and task-related operations: status queries, re-prioritisation, reassignment, exception handling. The everyday floor-supervisor interface, delivered as conversation.
Readiness note f42822
Setup. All three agents require Fusion AI Agent Studio configuration — FA IDCS OAuth2, SaaS app setup, and group assignment — before invocation. They are not toggle-on features; they are platform agents that need to be stood up and scoped per tenant.
Oracle’s Supply Chain Planning 26A release framed its AI agents around a specific idea: remove the routine coordination work from planners’ days so the judgement calls can get more attention. Three agents shipped.
Helps planning teams automate task coordination and improve planning cycle efficiency. Retrieves, assigns, and updates planning tasks within the planning workspace — the scheduling and chase work that most planners spend a surprising share of their week on.
Lets planners type simple business rules in conversational language to select orders for release. Oracle’s own example: “If releasing the order will delay any customer demand by more than seven days, hold it for review.” Natural-language rule authoring for order release policy.
Provides access to useful information on measure expressions and guides the planner through a simple chat experience. The metric-authoring layer most planning teams usually consult a power user for — now in a chat.
Pattern. All three are natural-language layers over existing planning machinery, not new planning algorithms. Oracle is not claiming to replace the planning engine — it is claiming to replace the friction of using it.
Procurement in 26A and 26B is where Oracle put the most AI-agent investment inside Fusion SCM. The logic is obvious: procurement work is already document-heavy, negotiation-heavy, and coordination-heavy — all the shapes LLM agents handle well.
Identifies, prepares, invites, negotiates, and awards bids autonomously. End-to-end sourcing run as an agent workflow, with humans in the loop for approvals and exceptions.
An agentic application from the Apr 9, 2026 batch. Coordinates the team of agents that surface high-priority sourcing decisions, contract context, and supplier risk into a single command surface.
Helps purchasing teams resolve external-price load errors. A utility agent — less marketed than Autonomous Sourcing, equally shipped.
Natural-language supplier search across historical transactions and supplier profiles, with a summary of how the agent interpreted the request. Transparent retrieval, not black-box.
In Self-Service Procurement 26B, preparers use this agent to create and submit requisitions through guided conversation.
A note on Contract Advisor. Procurement 26A revision history (19 Feb 2026): the Contract Advisor AI Agent was “Removed feature from update 26A.” Treat “AI-Driven Contract Review” as roadmap, not shipped.
On March 24, 2026 Oracle introduced Fusion Agentic Applications — 22 multi-agent applications across Finance, HR, SCM, and CX. On April 9, 2026 Oracle added 12 more for Finance and Supply Chain. These are not single agents; they are teams of agents coordinated by Agent Studio to own a full business workflow.
| Agentic Application | Function | What Oracle says it does |
|---|---|---|
| Design-to-Source Workspace | PLM & Procurement | Reduce product cost, cycle time, and compliance risk by shifting disconnected functions across engineering, supplier, and sourcing decisions into one coordinated process. |
| Product Readiness Workspace | PLM | Increase product launch efficiency and reduce delays; assess supply chain impact; boost compliance — launch readiness as a single guided workflow rather than manual tracking across teams. |
| Process Manufacturing Workspace | Manufacturing | Increase manufacturing quality, enhance issue detection, improve batch conformance, and reduce costs — moves production, quality, and cost analysis into one insight-driven workflow. |
| Production Shift Operations Workspace | Manufacturing | Streamline shift handoffs, reduce disruptions, improve output reliability with shift-readiness checks and carryover flagging. |
| Cost Accounting Close Workspace | Manufacturing & Inventory | Prioritise close work, reduce effort, and accelerate period close by surfacing material exceptions and next-best actions across manufacturing and inventory operations. |
| Warehouse Operations Workspace | Inventory & Logistics | Increase warehouse efficiency and expand insights across stock, inbound, outbound, and workforce — a supervisor-level agentic layer above WMS. |
| Maintenance Operations Workspace | Maintenance | Reduce unplanned downtime, speed up triage, and focus on highest-impact actions — maintenance ops as an agent-coordinated command centre. |
| Logistics Execution Command Center | Logistics / WMS + OTM | Minimise fulfilment disruption, identify urgent issues faster, unify data across transportation and warehouse operations. |
Pause — before we look at what else just shipped
Oracle isn’t selling AI features. They’re shipping an AI operating model.
The next three spreads cover the AI capabilities that sit alongside the conversational agents — the Predictive Fulfillment Dashboard that has been quietly shipping since 22D, the brand-new Logistics Execution Command Center agentic app released in 26B, and the platform plumbing that lets customers and partners build their own.
Conversational agents are the visible part of Fusion AI in Warehouse Management. Two more capabilities sit alongside them — one a quietly mature predictive feature shipping since 22D, the other brand-new in 26B.
An AI/ML feature that predicts Order Cycle Time, processing time and waiting time — identifying orders likely to miss expected service levels, shipping windows, or create warehouse bottlenecks.
Lineage: shipping since release 22D, continuously enhanced through the 25D online documentation and the 26A WMS Cloud AIML Guide. A real, production AI/ML feature — not a prototype.
An agentic app that unifies Warehouse Management and Oracle Transportation Management into one Command Summary — the “pulse” of execution across both. On launch, logistics managers see Orders at Risk and Priority Actions surfaced first; Ask Oracle takes natural-language queries for cross-system intelligence.
Citation: readiness note f44026 — the canonical 26B record.
A correction we’re grateful for. An earlier version of this magazine described the Predictive Dashboard as a “prototype.” Arun Murugan (VP, Software Development at Oracle and former CTO of LogFire) flagged that it has in fact been a full feature for some time — with the documentation due an update — and pointed us at f44026 for the new 26B Logistics Execution Command Center agentic app. Both corrections applied. The magazine is better for it.
Fusion WMS has been shipping AI/ML features since 22D — longer than the conversational-agent story implies. Oracle’s AI for Fusion SCM feature list names twelve WMS AI items across four release types.
Why this matters. The honest pitch is “four years of WMS AI/ML, plus a 2026 conversational and agentic-app surface” — not “AI just arrived in WMS in 26A.”
Announced March 24, 2026 as an expansion of Fusion AI Agent Studio. The Builder lets customers compose multi-agent workflows over Fusion data using natural language — without traditional coding. This is the bet Oracle is making: AI adoption in enterprise won’t be about “more Oracle-built agents.” It will be about customers building their own.
Historically, extending Fusion meant configuration first, then customisation with PaaS extensions and OIC integrations — a serious engineering exercise. The Builder offers a third path: if the workflow you want doesn’t exist as an Oracle-ship agent, you can compose it on the same platform the Oracle agents run on.
Implementation partners — including practices like ours that run Fusion and EBS implementations day-to-day — will increasingly deliver agents as a deliverable alongside configuration and integration. The skill mix this requires is not traditional functional consulting; it is prompt design, workflow composition, and agent evaluation.
Launched October 15, 2025. The Fusion Applications AI Agent Marketplace lets Fusion customers discover and deploy Oracle-validated, partner-built AI agents — the same platform the Oracle-ship agents use, with a third-party supply side alongside.
Oracle is positioning Fusion as an agent-friendly platform — not a closed AI store. The Marketplace, together with the Agentic Applications Builder, is how Oracle signals: the agents your business needs may not come from Oracle, and that’s by design.
The broader picture
Announced October 14, 2025 — one day before the Marketplace. The AI Data Platform is the data-layer foundation the agents (Oracle-built, customer-built, partner-built) all sit on: structured Fusion data, unstructured content, multimodal inputs, retrieval, vector indexes.
Agents without a data platform are demos. Oracle shipped the platform first, then opened the agent supply.
The cumulative platform story. Agent Studio (Mar 2025) → AI Data Platform (Oct 2025) → Agent Marketplace (Oct 2025) → Agentic Applications Builder (Mar 2026). Read in order, that’s a deliberate platform build — not a feature catch-up.
Not every workflow needs an agent. Not every buyer is ready to govern one. Honest buying conversations answer these five before the sales team reaches for a demo.
Read them together. Oracle has a real, named, shipping AI story. Adoption requires platform discipline, governance ownership, and a clear posture on build-vs-buy. Without those, the agents are demos. With them, they are operational leverage.
In Closing
Per Oracle’s May 2026 AI Features Report: 178 SCM AI items, including 54 named agents and 10 agentic applications in the 26A and 26B releases alone — on top of a developer studio, data platform, marketplace, and agentic-applications builder shipped between March 2025 and April 2026. Agents surface through a small set of consistent invocation patterns: Home with Ask Oracle, in-page Ask/ellipsis, and product-specific Ask AI menus.
This is not the AI marketing you’re used to from enterprise software vendors. It is a coherent product strategy, delivered publicly, on a schedule — and it deserves to be read the way Oracle wrote it. Read the release notes. Ask the governance questions. Pick a posture on build-versus-buy.
For Vol. 07. Sales Order Command Center · Purchase Order Status Advisor · OTM Bulk Plan Diagnostic Analyst · OTM Rate Inquiry Assistant · Change Order Auditor · Work Order Completion Assistant · Inventory Task Allocation Assistant · plus the 16 agents in Inventory Management 26A–26B.
Sources, verbatim
Oracle News — AI Agent Studio announcement (Mar 20, 2025)
Oracle News — AI World 2025: Agent Marketplace & AI Data Platform (Oct 14–15, 2025)
Oracle News — AI Agents for Supply Chain (Feb 10, 2026)
Oracle News — Fusion Agentic Applications & Builder (Mar 24, 2026)
Oracle News — 12 more Fusion Agentic Applications for Finance & SCM (Apr 9, 2026)
Oracle Fusion Cloud WMS 26B What’s New (docs.oracle.com)
Oracle 26A readiness notes f42424 / f42576 / f42822 (platform note f42925)
Oracle Fusion Cloud Procurement 26A / 26B What’s New
Oracle SCM Blog — 26A Supply Chain Planning, Procurement posts
Oracle AI for Fusion SCM feature list · Oracle AI Features Report (May 2026) — primary inventories