Published: July 25, 2026, 6:15 PM IST
Oracle opened AI Agent Studio for Fusion Applications to pro-code developers on July 14, 2026, letting engineers build, test, and deploy Fusion Agentic Applications using VS Code, Claude Code, Codex, Git, and standard command-line tools. The update completes Oracle’s three-tier builder stack — natural-language no-code, low-code, and now pro-code — for autonomous enterprise software that runs inside Oracle Fusion Cloud Applications. Oracle offers the capability at no additional cost to existing Fusion Applications customers.
The announcement adds professional developers to a platform Oracle says already carries 80,000 certified experts, more than 1,000 delivered AI agents, and 22 Fusion Agentic Applications launched earlier in 2026. Chris Leone, Executive Vice President of Applications Development at Oracle, framed the shift as software moving “beyond systems that record work to systems that actively drive and execute outcomes.” Accenture, Deloitte, and PwC back the rollout as implementation partners.
What is Oracle AI Agent Studio for enterprise developers?
Oracle AI Agent Studio is a builder platform that lets enterprise developers create AI agents which run natively inside Oracle Fusion Applications, reusing the business objects, workflows, security, approvals, and audit trails that already exist in the system. The July 14 update extends this platform to pro-code developers, who can now assemble Fusion Agentic Applications with professional coding tools rather than only natural language or low-code interfaces. Oracle also announced a public GitHub repository that will supply templates, starter projects, and sample applications for these developers.
Fusion Agentic Applications are teams of specialized agents that act as autonomous software targeting a defined business outcome, such as accelerating financial close, improving collections, reducing service escalations, optimizing workforce operations, and streamlining supply chain execution. Natalia Rachelson, Senior Vice President of Applications Development at Oracle, described a renewal-management example in which an agent identifies opportunities, drafts contracts, and sends them while tracing every decision against Fusion records.
Why does building with Claude Code and Codex matter?
Support for Claude Code and Codex means developers can use AI coding assistants to build AI agents, creating an “AI building AI” loop that accelerates enterprise agentic deployments. Oracle listed VS Code, standard CLIs, Git-based workflows, Claude Code, and Codex as the supported pro-code tools, and said these builders operate under the same governance, security, and auditability controls Oracle applies to its own agents. This positioning aligns Oracle with the broader movement toward AI agents handling business tasks across corporate software.
Oracle’s core argument is architectural: agents that live inside the enterprise system act on real transactional data with real governance, avoiding the “bolt-on AI” problem where identity, approvals, and audit trails must be rebuilt outside the system. Holger Mueller, Vice President and Principal Analyst at Constellation Research, said Oracle “keeps agents, security, APIs, access, and governance all within a well-defined, trusted, and proven modern application platform,” in contrast to approaches that build agents outside the application platform.
How does Oracle compare to rival agent platforms?
Oracle differentiates its agent strategy through a native runtime inside Fusion, which removes the separate infrastructure, identity management, and audit trail that external agent platforms require. Competing enterprise agent strategies include SAP Joule, Microsoft Copilot for Dynamics 365, and Salesforce Agentforce, each of which layers agents onto existing systems. Oracle also positions Fusion Agentic Applications against lightweight external orchestration tools, such as Zapier AI, Make.com, and n8n, by arguing that governance — not model quality — decides enterprise AI outcomes. This governance-first design mirrors emerging patterns for AI agents accessing enterprise tools under tight identity controls.
WithO2 tracks this shift on our hub covering the best AI agents for business tasks, where Oracle’s Fusion Agentic Applications now sit alongside Agentforce, Copilot, and standalone agent builders.
For Context
WithO2 has covered the expansion of autonomous agents into daily business software, including OpenAI’s ChatGPT for Work and 1Password’s password-free agent access for Claude. Oracle’s pro-code release continues the same trajectory: agents moving from assistants into governed, transaction-executing software.
Our Take
Oracle is betting that enterprise AI wins or loses on governance, not on model quality. By forcing every agent through Fusion’s security layer, Oracle wagers that enterprises will pay a premium for auditability over the flexibility of external agent platforms. The risk is real: if companies decide lightweight external orchestration is good enough, Oracle’s tight coupling becomes a moat no one wants. The 80,000 certified experts are the strongest asset here — that trained channel is what will push the product into production.
Frequently asked questions
Does Oracle AI Agent Studio cost extra for Fusion customers?
No. Oracle offers AI Agent Studio at no additional cost to existing Oracle Fusion Applications customers, according to Oracle’s July 14, 2026 announcement.
Which developer tools does the pro-code builder support?
Oracle’s pro-code builder supports 5 tool categories: VS Code, standard command-line interfaces, Git-based workflows, Claude Code, and Codex.
What are Fusion Agentic Applications?
Fusion Agentic Applications are teams of specialized AI agents that act as autonomous software inside Oracle Fusion Applications, targeting defined outcomes such as financial close, collections, service escalations, workforce operations, and supply chain execution.
Sources: Oracle press release, July 14, 2026; SiliconANGLE, July 14, 2026.
