Oracle is one of the world’s largest enterprise software companies.
Its products run the back offices of the major enterprises you recognise.
Think ERP systems that manage corporate finances, HR platforms that handle payroll and hiring, supply chain software that moves goods across continents, and cloud databases that store the data enterprises run on.
Beyond software, Oracle operates Oracle Cloud Infrastructure (OCI), which competes with AWS, Azure, and Google Cloud for AI workloads.
The enterprise serves hundreds of thousands of customers across 175 countries and employs over 150,000.
In today’s AI at the Top, we’ll learn how Oracle approaches AI and considers itself a client zero to test the Oracle AI it builds for its clients.
Oracle has no CAIO or AI Council
Some enterprises we have studied in our newsletter have a governance structure.
J&J ran a central governance board across 900 experiments before handing ownership to business units.
Bank of America filters every use case through a 16-pillar framework.
Mastercard runs a hub-and-spoke model with a Chief AI and Data Officer at the centre.
Oracle leaned towards an executive-led approach.
CTO and Chairman, Ellison sets the thesis. CEOs Clay Magouyrk (infrastructure) and Mike Sicilia (applications) execute it. CIO Jae Evans oversees internal IT adoption.
The leadership has been closely involved in technology or has led tech teams for decades. This makes it no brainer for executives to lead the enterprise's AI strategy.
On the model side, Oracle takes the multi-vendor route.
Its internal AI agent studio supports OpenAI, Anthropic, Cohere, Google, Meta, and xAI, routing tasks to whichever model performs best.
Goldman Sachs uses a similar architecture for the same reason: no single-vendor lock-in, faster integration when better models ship.
Enjoying this read? Share the smartest part with your followers. Click to Share.
YOU DECIDE
🌟 Want to be featured in the next issue? Reach out with your best AI use case and we’ll spotlight it.
🏢 What company do you want us to cover next?
How Oracle uses AI to build its own software
Of course, engineering is one of the clearest proof points of AI applications for multiple enterprises.
Pick any company we covered. AI applications in coding are a given, irrespective of the industry.
Oracle is no different. It understands how Oracle AI is crucial for their customers, so they treat themselves as client zero to stress-test their tools internally:
Oracle Code Assist, built on OCI Generative AI, is used by thousands of Oracle developers for code generation, debugging, dependency upgrades, and language migrations.
A companion tool, Code Agent, runs inside Oracle’s CI/CD pipeline, the automated system that tests and deploys code, doing reviews that previously required senior engineers.
For application development, Oracle leans on APEX AI Code Generator. Devs express intent and the system writes the procedure.
In Q3 FY2026, smaller engineering teams shipped three new CX applications (lead qualification, sales orchestration, automated selling) plus a website-generation agent.
It’s good to see Oracle validate internal AI use through shipping a product rather than publishing benchmarks like “15% of our code is written by AI”.
Beyond coding, Oracle stress-tests its AI internally across multiple domains and teams:
What’s next for Oracle
In March 2026, Oracle reported a restructuring charge of up to $2.1 billion to shift towards AI-driven development.
On the product side, fusion agentic applications went generally available in April 2026. They are Oracle’s autonomous, agentic workspaces across finance and the supply chain, covering payables, ledger, planning, and payments.
Oracle has been running these on its own books for months, so customers get a system that has already been stress-tested at enterprise scale.
Also, the company is on the path to add agentic capabilities across all four fusion pillars:
Enterprise resource planning
Human capital management
Supply chain management
Customer experience
Oracle wants to build fewer standalone tools and embed more agents directly into existing employee workflows.
This ensures fast and high adoption because it’s already inside the system they open every morning.
What Enterprise Leaders Can Learn from Oracle's AI Strategy
Deploy internally first, sell what survives. If you are your own customer, it’s one of the strongest ways to iterate on the product before you GTM. Strong teams let internal use set the quality bar.
Where does your AI productivity eventually land? Map your productivity gains to impactful results. Hours saved, automations, and reducing workforce are okay, but features, revenue, etc., make better metrics.
Make it easy to switch models. AI moves fast. Today’s hero model gets to second place in just a matter of months. But your business always needs the best model, so optimising architecture for a multi-vendor approach seems wise, even if it costs more or takes time.
Governance is a strategic choice. Companies like BofA, Mastercard, and J&J have councils to monitor AI progress. Oracle chose to become executive-led. Multiple factors influence your AI governance policy. Case studies like today make you aware of industries and applications, so you can only choose what’s best for you.
There is no one way to win the game.
Sources:
YOUR TURN