Release 26B and Agentic Finance: Oracle Puts AI Agents Inside GL

Title: Release 26B and Agentic Finance: Oracle Puts AI Agents Inside the General Ledger

Date: 19 May 2026

Type: Blog

Author: Michael Hulbert (michael@saasiq.ai)

Word count: 1,060

Reading time: 5 minutes

Published: 19 May 2026

Tags: Oracle Fusion, Release 26B, Agentic ERP, AI Agents, Oracle Financials, Source-to-Settle

Oracle Release 26B began rolling out in mid-May 2026 with 22 new Fusion Agentic Applications spanning ERP, HCM, SCM, and CX. The headline for finance teams is a pair of purpose-built agents, the Source-to-Settle Assurance Advisor and the Record-to-Report Assurance Advisor, that operate autonomously inside transactional workflows. These are not copilots sitting alongside a user. They are native agents embedded at the application layer, executing within the financial close and procurement cycles that run the business. This piece examines what 26B means for Oracle ERP customers, why agentic finance changes the operating model, and how the broader ecosystem is aligning around this shift.

What 26B Actually Ships

Release 26B delivers 22 Fusion Agentic Applications across the full Oracle Cloud Applications suite. The rollout window runs from mid-May through mid-August 2026, which means most Fusion customers will receive the update during their next quarterly maintenance cycle. Within ERP specifically, the two standout additions are the Source-to-Settle Assurance Advisor and the Record-to-Report Assurance Advisor, joined by a new Access Request Assistant for provisioning workflows.

The Source-to-Settle Assurance Advisor monitors the full procurement-to-payment cycle, identifying exceptions, reconciliation gaps, and compliance risks without waiting for a human to run a report. The Record-to-Report Assurance Advisor does the same for the financial close, scanning journal entries, intercompany transactions, and period-end reconciliations. Both agents operate continuously rather than on a batch schedule, which represents a fundamental change in how ERP quality assurance works.

Agents, Not Copilots

The distinction between a copilot and an agent matters enormously for finance organisations. A copilot suggests actions to a human who then decides whether to proceed. An agent acts within defined parameters, executing decisions, flagging exceptions, and routing outcomes to the appropriate approver. Oracle has been explicit that these 26B agents are the latter category, designed to operate autonomously within the boundaries of configured business rules.

We see this as Oracle's answer to the question that has hung over enterprise AI for the past two years: where does automation stop and autonomy begin. With 26B, Oracle is placing that boundary inside the financial close and the procure-to-pay cycle, two of the most audit-sensitive processes in any organisation. The implications for internal controls, segregation of duties, and continuous monitoring are significant, and we will return to those in a moment.

Fifty-Plus Agents and the Agentic Application Builder

Beyond the named agents in 26B, Oracle now offers more than 50 role-based AI agents embedded directly into Fusion Cloud. These span finance, procurement, HR, supply chain, and customer experience functions. Oracle has branded this broader initiative as Agentic Finance, signalling that the company views autonomous agents as the default operating model for its cloud applications, not an optional add-on.

The Agentic Application Builder, also introduced this cycle, allows customers to create their own outcome-focused agentic applications without writing code. This is Oracle's platform play: rather than limiting agents to what Oracle ships out of the box, the builder gives customers and partners the ability to construct agents tailored to their specific processes. For organisations with complex, multi-entity structures or industry-specific workflows, this is where the real value may sit over time.

Ecosystem Signals Confirm the Direction

Several developments this week reinforce the scale of Oracle's agentic commitment. IBM and Oracle announced an expansion of their 40-year partnership, focused on unifying procurement and supply chain data with AI agents. This is not a press release partnership. IBM's involvement in procurement intelligence, combined with Oracle's agent framework, points to a practical integration layer that will affect large-scale Fusion deployments.

Eightfold AI launched an agentic interview intelligence integration with Oracle Fusion Cloud Recruiting, running natively on OCI.

IHH Healthcare, one of the largest private healthcare groups globally, confirmed it is consolidating its enterprise backbone on Oracle Fusion Cloud Applications. Tamaki Health in New Zealand deployed Fusion Cloud HCM and reported reducing its hiring cycle from months to three weeks. Each of these signals points to the same conclusion: the ecosystem is building around Oracle's agentic architecture, not waiting for it to mature.

Wall Street Agrees

Wedbush raised its Oracle price target twice in the space of three weeks during May, citing 84 percent cloud growth and a reported AI infrastructure backlog of 553 billion US dollars. Analyst coverage has shifted from questioning whether Oracle can compete in AI to assessing how quickly the agentic applications will drive incremental cloud consumption. The financial narrative is now aligned with the product roadmap, which is not always the case with enterprise software vendors.

For customers evaluating Oracle's long-term commitment to the agentic model, the investor signals matter. Capital follows conviction, and Oracle is directing both internal investment and external messaging squarely at agentic applications as the growth driver for the next several years.

What This Means for Control Frameworks

Agentic finance introduces a new category of control risk that most organisations have not yet addressed. When an agent continuously monitors journal entries or procurement transactions, it holds persistent access to sensitive financial data and can initiate actions based on its analysis. Traditional audit approaches that sample transactions quarterly are not designed to govern an agent that processes thousands of records per hour.

Organisations running Oracle Fusion ERP need to update their control frameworks to account for agent identities, agent-specific segregation of duties policies, and real-time exception routing. Oracle published its five-step cybersecurity framework for agentic ERP earlier this month, and we strongly recommend using that as the baseline for a governance assessment before 26B lands in your production environment.

The SaaSiQ Perspective

At SaaSiQ, we view 26B as the release where agentic ERP stops being a concept and becomes an operational reality. The Source-to-Settle and Record-to-Report Assurance Advisors are not experimental features tucked into an innovation sandbox. They are production-grade agents embedded in the core financial processes that CFOs and controllers are accountable for.

Our recommendation to every Oracle Fusion ERP customer is straightforward: treat 26B as a governance event, not just a feature update. Map the new agents against your existing control framework, confirm your segregation of duties policies account for agent identities, and establish exception routing before the agents go live. The organisations that prepare now will capture the productivity benefits of agentic finance without the audit findings that come from governing it after the fact.

Copyright SaaSiQ.ai, 2026. All rights reserved.