Building a pragmatic framework for scalable and trusted AI

AI continues to dominate headlines, board agendas and investor updates. The pressure to act is intense. Yet inside organizations, that mandate often turns into hesitation: Where do we start? How do we invest wisely without being left behind?

On paper, AI promises sweeping transformations — automating full functions, radically reducing costs and accelerating decisions. In practice, the current reality is more nuanced. While AI has advanced considerably, delivering sophisticated generative, predictive and analytical capabilities, its most widely adopted applications often focus on task-level enhancements: drafting performance goals, writing first-pass job postings, or summarizing call notes. These contributions, though undeniably helpful, highlight a persistent gap between AI’s transformative potential and tangible, widespread organizational impact. This discrepancy leaves executives navigating three pressing hurdles:

Confusion

Teams are told to ‘use AI,’ with little clarity on where to begin or what success looks like.
Cost

Building custom AI agents from scratch demands expensive infrastructure, clean data and scarce talent. This locks many leaders out by price and complexity.
Caution

Even where interest is strong, adoption slows with concerns for governance, security and potential unintended consequences.

Leaders don’t want AI for the sake of AI. They want measurable and meaningful outcomes that align with their organization’s goals, delivered responsibly and without incurring unnecessary risk. That’s where a balanced, strategic framework becomes vital. KPMG professionals are helping organizations leverage Oracle AI Agent Studio for Fusion Applications to bridge this gap, providing a pragmatic pathway to agentic AI: a low-code environment to build, configure and scale AI agents. Unlike costly custom builds or lightweight embedded features that only address the margins of workflows, Oracle AI Agent Studio creates a way to move forward responsibly, strategically and with confidence.

Our AI strategy is all about embedding artificial intelligence where it matters most — directly into the applications and workflows that our customers use every day. The Oracle AI Agent Studio gives customers the freedom to innovate while still leveraging Oracle’s secure infrastructure and AI services. It’s about putting powerful tools in the hands of business users, not just data scientists.

Kaushal Kurapati

Group Vice President,

Fusion AI, Oracle

From automation’s limits to augmentation’s potential

The first wave of generative AI centered on automation: machines replacing repetitive or rule-based tasks. It delivered efficiency gains — a finance team saves hours by automating reconciling processes, an HR team lightens its administrative load. Many of these wins were siloed. The broader function, let alone the enterprise, didn’t fundamentally change.

Executives today want (and need) more. They want augmentation: AI that enables human performance, is embedded directly into workflows, and can evolve business models and ecosystems — all to uplift human potential and foster lasting value.

Consider a premium beef producer in Australia — part of an industry that’s challenged with high initial investments, long breeding processes and market volatility. For years, the company collected animal health and feed data, but struggled to turn it into commercial outcomes. By turning on AI agents in Oracle Fusion Cloud Applications and leveraging its machine learning capabilities, KPMG professionals helped the beef producer identify seven key variables that predicted beef marbling quality up to two years in advance. The outcome was a sales strategy that guaranteed demand and reduced exposure to market swings — a critical advantage in an industry where every decision carries heavy costs and long-term risks.

Or a UK-based media organization. Month-end close was dominated by manual searches through thousands of spreadsheet rows for anomalies. With AI agents in Oracle Fusion Applications, KPMG professionals were able to help the finance team start each cycle with a shortlist of flagged items with the context attached. Instead of chasing data, the team is able to investigate, resolve and report with greater speed and accuracy.

In both cases, augmentation didn’t just save time, it shifted effort from low-value mechanics to high-value judgment. Farmers secured stronger margins and finance teams improved reporting confidence. That’s the compound impact of humans and AI working together.

AI agents aren’t replacing our jobs — they’re assisting us in them. AI has no real judgment of its own, so safeguards and human oversight remain critical. As AI becomes a natural part of business, we’ll move beyond automating today’s tasks to rethinking how work itself is done and how roles are defined.

Linda Chai

Partner, KPMG Enterprise

KPMG Australia

Building trust, embedding governance

For every executive eager to seize the possibilities of AI, there are others in the room who are hesitant. This tension can stall progress. The breakthrough often comes from starting small — pinpointing the right opportunities that can deliver value within a few weeks, not years.

That’s how a US-based gaming and entertainment company gained momentum. The company’s AI steering committee was energized about the potential of AI agents, but those who were less advanced in their understanding of the technology remained skeptical. KPMG teams worked with the company to pilot two targeted use cases: one to answer procurement policy queries and the other to flag anomalies in payables.

The impact was immediate. Employees who had been weighed down by repetitive tasks saw relief, trust in the technology grew — and most importantly — the quick wins sparked a pipeline of new ideas to use AI agents across the business.

AI is moving so fast that it’s hard for organizations to keep up. We help break it down, so clients can take the first steps. The key is not to be apprehensive. Instead, start small, lean on trusted advisors and build momentum from there.

Nick Yoko

Managing Director, Advisory

KPMG in the US

But momentum without governance is fragile. A UK government department illustrates the other side of the equation. Leaders knew AI agents could transform its HR operations, but rolling out AI agents across a sprawling, multi-entity structure raised governance questions. The KPMG team helped embed AI directly into process maps, trained managers on responsible use, and built accountability frameworks for testing and monitoring. What could have been a rollout mired in uncertainty became one built on clarity and confidence.

The lesson? Quick wins build trust and governance sustains it. Both are required to scale.

You can’t just bolt AI onto existing processes. Without industry context and governance designed from the start, you risk embedding errors and bias in the system that doesn’t work within the sector rules, constraints and risks – and AI agents become clever tools in search of problems.

Ambreen Gul

Director

KPMG in the UK

Practical steps for agentic AI adoption

Jumping into AI without a plan leads to wasted spend and half-built pilots. The most successful leaders set the direction first, then move with focus and discipline. Here’s how to start:

Anchor AI agents in strategy, not hype

Define what matters most — faster close cycles, sharper talent decisions, stronger customers responsiveness for example. Then, set KPIs and guardrails that can keep AI agents aligned.
Start with what you already own

Many organizations already pay for embedded Oracle AI features. Turning these on can surface quick wins without major investments.
Experiment safely

Proof-of-concept and sandbox environments create the space you need to test, learn and adapt without risking core operations.
Build governance in from the start

Bias, security gaps and misaligned decisions are real risks. Responsible use policies, monitoring frameworks and transparency protocols build trust in both the technology and its outcomes.
Move in sprints

Incremental rollouts with human checkpoints make progress tangible, mitigate risks and build confidence at each stage.
Co-evolve with the platform

Oracle’s quarterly updates mean the tools keep advancing. Leaders should evolve business processes in step to ensure their enterprise is never left behind.

Beyond today’s agent pilots

AI agents aren’t the finish line — they’re the starting point for a more profound shift on how work will get done. Today, they streamline tasks and sharpen decisions. Tomorrow, they’ll reshape how functions like finance, HR and procurement are designed, overlaying human judgment onto machine precision.

To get here, today’s leaders don’t have to rush ahead recklessly. They also can’t stand still. Instead, they can embrace the middle ground. Oracle AI Agent Studio lowers the barriers to adoption, but the real advantage is in how you use it: starting small, embedding safeguards and scaling with discipline.

How KPMG can help

KPMG professionals can help you unlock the potential of agentic AI for your organization with Oracle AI Agent Studio.

KPMG firms’ extensive experience with Oracle and AI integration, along with responsible AI and our enable-embed-evolve framework, can help you prioritize investments, align initiatives with business goals and guide you from strategy development and implementation through to transformation. KPMG professionals bring the industry expertise you need to customize AI and help meet the needs of your unique business, while helping to generate lasting value for your organization.