For decades, there’s been a disconnect in enterprise software. The people who understand the work best—the supply chain manager who can spot inefficiency in seconds, the finance analyst who sees a faster way to close the books, the HR lead who can picture a smarter scheduling tool—are rarely the people equipped to build the systems that fix what they see. The people who can build those systems, meanwhile, aren’t living inside the day-to-day workflow. The result is a familiar frustration: Domain experts file a ticket, and then they wait.
That disconnect is starting to disappear. Eighty percent of low-code users will come from outside IT departments, up from 60 percent in 2021—a sign that the people building enterprise software are increasingly the people who use it, and not just professional developers.
Where AI Deployments Stall
The problem is that this transition toward empowering more builders hasn’t solved the harder question of how to do it safely and at scale. Most enterprises have big AI ambitions, but those efforts often stall when AI tools get built in isolation, outside the core systems where the business actually runs. Every deployment built that way often ends up carrying a second, hidden project: retrofitting enterprise controls after the fact.
To avoid creating a governance problem, the building needs to happen inside a platform that already has security and controls built in, rather than in a separate environment that needs fixing later. It’s this structural mismatch between how fast AI moves and how centralized development models are built to operate that gets in the way of organizations trying to deploy new technology.
The Builder Economy
The demand to expand who can build is growing faster than the controls and safety mechanisms needed to allow it, which is increasing the need for a new era of enterprise technology.
Oracle sees this as the beginning of a new app-builder economy for enterprise AI. For example, the AI-powered builder experience in Oracle AI Agent Studio for Fusion Applications empowers business users, enterprise developers, and systems integrators to turn domain knowledge into AI automations and agentic applications with built-in security, governance controls, and auditability. In the process, it’s helping AI meet employees where they are: employees who think in natural language, operations leads who know their workflows cold, and developers who want to move faster with the tools they already know without losing control.
Building Inside Enterprise Guardrails
The open question for organizations adopting AI is where the governance, security, and controls integrate with the development process. Should it be built into the platform from the beginning or added later as separate layers? Oracle’s approach is to provide a complete development platform inside its Oracle Fusion Applications environment that organizations already use to run their business so governance and security exist from day one.
This builder experience in Oracle AI Agent Studio is meant to empower both those who have never coded before as well as professional developers. The latter can still work in familiar tools—VS Code, Git, Claude Code, and OpenAI Codex—while remaining inside that governed framework. A public repository supplies templates and reference architectures to speed up builds. Right now, Oracle offers more than 1,000 AI agents and over 22 Fusion Agentic Applications that customers and partners can leverage to build out new capabilities in a Fusion-native environment.
From Systems of Record to Systems of Outcomes
Fusion Agentic Applications mark a new category of enterprise software: applications built around outcomes and powered by groups of specialized AI agents that can reason through a problem, coordinate with one another, make decisions, and then take action. This capability sets them apart from standalone agents, copilots, and disconnected automation tools as they operate inside the enterprise systems where work is already happening.
When domain experts can build agentic applications natively inside the systems they already use, work happens faster. The underlying systems of record start to function as systems of outcomes: accelerating financial close, reducing service escalations, optimizing workforce scheduling, and streamlining supply chain execution. Reusable agents, workflows, and templates compound that advantage over time, helping organizations move faster on each subsequent build. The net result is far more than just efficiency. It allows organizations to begin rethinking how work gets done by capturing operational knowledge in applications that can be securely built, reused, and continuously improved.
The early data suggests this approach pays off. Organizations that successfully move AI from pilot projects to production-scale processes report an average ROI of 1.7x, with cost savings of 26 to 31 percent across supply chain, finance, and operations functions. As the builder economy expands—and projections suggest it will—the advantage is likely to go to organizations that enable more employees to build, while keeping development inside platforms with built-in governance, security, and auditability.
