Oracle has publicized its intent to join the GenAI party with a new AI-based tool in the works, Oracle Code Assist, which will help developers build applications faster.
While the technology looks promising, particularly for Java and SQL developers, it is perceived as late to the party, given the lead that tools like GitHub Copilot, OpenAI’s ChatGPT and others have. Moreover, Oracle Code Assist is not even available outside of Oracle yet.
Fashionably Late?
“Oracle comes late to the party, but those are sometimes the best guests — as they can help replenish the party,” said Holger Mueller, an analyst at Constellation Research. “And in the case of Oracle, they are bringing a lot of GPU capacity, a lot of Java domain expertise and a lot of experience in coding in general to the party. So to some extent, they get the party started.”
Brad Shimmin, an analyst at Omdia agrees.
“Yes, they are comparatively late if you’re talking about generalized copilot in the style of Microsoft Copilot, CodeLlama, et al.,” he said.
However, “Here we’re talking about a code generation system that will likely be fine-tuned to the Java language and pre-trained upon some very high-quality code,” Shimmin said. “I would expect that their efforts here will mirror the work the company is doing with SQL generation, which is specific to the Oracle software development paradigm and more importantly aligned with and grounded in the customer’s own context.”
And more than that, Oracle is building this on top of a rich legacy of code optimization algorithms and techniques that the company has gleaned over many decades in optimizing customer database queries, he explained.
The announcement of Oracle Code Assist is intended to signal to customers and developers that Oracle is actively working on AI-assisted development tools tailored to their specific needs and expertise.
Potential Widespread Adoption
“Developers will be able to use Oracle Code Assist to quickly upgrade entire Java applications to newer versions. This ability to rapidly upgrade Java using a coding assistant significantly increases the ability of enterprises to improve application resilience and performance, incorporate the latest JDK features and bolster their security posture,” said Arnal Dayaratna research vice president, IDC, in a statement. “Oracle Code Assist can also accelerate database upgrades with respect to the refactoring and testing of complex and often customized PL/SQL code, which can be time-consuming and quite expensive. Given the ubiquity of Java and databases in the enterprise, Oracle Code Assist is likely to achieve widespread adoption by prominent organizations globally.”
Oracle Code Assist is an AI code companion that boosts developer velocity, enhances code consistency, and is optimized for specific scenarios like Java applications and Oracle Cloud Infrastructure (OCI), Vijay Kumar, vice president of product marketing and application development for OCI, told The New Stack.
Key Capabilities
Oracle Code Assist is powered by large language models (LLMs) running on OCI and optimized for application development on OCI. It will provide context-specific suggestions tailored to an organization’s best practices and codebases, and can be used to update, upgrade, and refactor code written in most modern programming languages such as Python and others, Kumar said.
The tool is a plugin that developers can install in popular development environments like JetBrains IDEA IntelliJ and Visual Studio Code, with more to come in the future.
The tool is powered by AI models running on Oracle Cloud Infrastructure, trained and fine-tuned for Java, SQL, and NetSuite’s SuiteScript.
Key features include code generation, test case generation, documentation creation, automatic language upgrades, code analysis for optimization, and code explanation for new talent.
Currently, Oracle Code Assist is only available to a subset of the company’s 40K+ internal developers and will be released to all of them in the very near future., Kumar said.
“Custom applications are the method by which every organization differentiates itself globally, but building, deploying, and maintaining applications is labor and time intensive,” said Aanand Krishnan, vice president, product management, OCI, in a statement. “Organizations leveraging Oracle Code Assist can help developers increase velocity and code consistency to improve the long-term maintenance of applications in a safe, rigorous, and compliant manner.”
Meanwhile, differentiators from other AI coding tools include targeting specific use cases important to Oracle’s customer base, leveraging Oracle’s expertise in languages like Java and SQL, and offering additional non-LLM-based features like bug fixing and performance tuning.
In addition, the tool helps with language upgrades, refactoring old code, and providing suggestions for improved performance.
Early internal usage patterns show developers primarily using the tool for coding suggestions, documentation, and testing, Kumar said.
Oracle is continuously evaluating and fine-tuning AI models to provide the best performance and experience for their developer base, he said.
The post Oracle’s Code Assist: Fashionably Late to the GenAI Party appeared first on The New Stack.
Oracle has a new GenAI tool brewing internally that will hit the market in the future to tap a sweet spot for Java and SQL developers, among others.