If you manage an Enterprise Planning and Budgeting Cloud Service (EPBCS) environment, your spring just got a whole lot busier—and a lot more exciting.
After a brief pause in monthly updates to thoroughly iron out Essbase 21c backend fixes, Oracle officially roared back with its Spring 2026 EPM Cloud updates. This quarter’s drops are heavy on productivity, offering a massive leap forward in Generative AI, smoother integration pipelines, and critical infrastructure updates that you need to account for immediately.
Let's break down the most impactful new features hitting your EPBCS environment this quarter and what they mean for your finance team.
1. Agentic AI & Generative AI Go Global
Oracle isn't just treating AI like a buzzword; they are embedding it directly into the fabric of FP&A.
Global Gen AI Availability: Generative AI capabilities have officially rolled out across all regions in the primary OC1 realm, as well as OC4, OC19, OC29, and OC42.
Agentic AI REST APIs: Oracle introduced three new REST APIs specifically designed to fuel Agentic AIinteractions. This includes a new
Get Application SummaryAPI that generates an AI-focused brief of an EPM application, alongside advanced JSON data grid exports. If you are building internal AI agents or automated bots to pull or analyze planning data, these APIs are your new best friend.
2. Infrastructure & Scripting: The Mandatory Groovy Update
This is the "stop what you're doing and check your code" headline of the quarter.
Oracle has deployed a mandatory Groovy engine upgrade. This update introduces much stricter validation rules for your custom calculations and rules. If you have complex, tailored Groovy scripts running your driver-based planning or data movements, you need to test them in your sandbox environment immediately to prevent scripts from breaking in production.
On the bright side, Groovy rules have also been enhanced to natively support the copying of relational data, comments, attachments, and supporting details, giving you far more automation flexibility.
3. Data Integration: Essbase Load Rules & Smarter Pipelines
Data integration admins are getting some highly requested quality-of-life improvements this quarter.
Native Essbase Load Rules Support: In a massive win for hybrid or migrating environments, you can now migrate classic Essbase data load rules directly into Cloud EPM applications without completely rebuilding them as Data Integrations from scratch.
Restart Pipelines from Failed Stages: We’ve all been there: a massive data pipeline fails at step 7 out of 10, and you have to rerun the entire thing from midnight. No more. You can now restart automated pipelines directly from the specific stage that failed.
Pipeline Wait Times: You can now inject precise wait times and pauses within pipeline jobs, allowing dependent backend processes or file transfers time to breathe before the next job triggers.
4. Next-Level Security & Maintenance: Physical Backups
Historically, EPBCS administrators relied solely on Artifact Snapshots for daily maintenance. This quarter, Oracle introduced a brand-new Environment Backup system.
This creates a full backup at the physical tier during daily maintenance. Alongside this, the new restoreEnvironment EPM Automate command allows you to roll back and restore environments with a much deeper layer of data integrity.
Additionally, the essbaseBlockAnalysisReport
Head's Up: Watch Your EPM Automate Scripts!
A minor change that could cause a major headache: the default installation path for EPM Automate on Windows has moved to the standard Program Files\Oracle\EPM Automate directory. If you have batch files or Windows Task Scheduler scripts hardcoded to legacy pathing, you'll want to update those file paths before your next scheduled run.
The Bottom Line
The Spring 2026 updates prove that Oracle is laser-focused on turning EPBCS into a highly automated, AI-driven machine. From Agentic AI capabilities to granular pipeline controls, these changes remove friction for finance teams—provided your admins proactively test the new Groovy validation limits.
