New updates to Konveyor AI: Use AI-driven application modernization without fine-tuning a model

Last year, Red Hat engineers introduced generative artificial intelligence (gen AI) to Konveyor, an open source community project aimed at helping organizations modernize applications for cloud-native environments. Today, we’re pleased to announce that the Konveyor community released the 0.1 version of Konveyor AI, which integrates gen AI into the migration workflow. With the development of Konveyor AI (Kai), Red Hat has been working together with the Konveyor community to integrate the experience of large language models (LLMs) into application modernization. By enhancing the existing migration toolkit for applications, Kai aims to streamline re-platforming and refactoring efforts, making modernization more accessible and cost-effective. 

Bringing AI-powered modernization to developers

Kai integrates gen AI with static code analysis to accelerate modernization tasks within a developer’s IDE. By combining LLMs with structured migration data, developers can automate source code suggestions and changes, gain insights from past transformations and reduce the manual effort required for large-scale migrations. By using retrieval-augmented generation (RAG), Kai enhances LLM outputs with historical code changes and analysis data, meaning modernization tasks are context-specific. This allows organizations to achieve AI-driven modernization without requiring model fine-tuning, making it a powerful tool for large-scale transformations.

New updates in Konveyor AI

The 0.1 release of Konveyor AI builds on this foundation with several key enhancements:

• Static code analysis: Identifies potential issues when adopting new technologies (e.g., modern Java frameworks, adopting containers).
• Issue tracking: Keeps a history of resolved modernization challenges.
• Enhanced migration intelligence: Encompassing 2,400 predefined rules for different migration paths with the option for users to define custom rules for unique scenarios.
• VS Code extension: Provides suggested code changes within the IDE.
• Model-agnostic AI: Konveyor does not lock users into a specific LLM, enabling flexibility in AI-driven modernization.
• Agentic AI for intelligent responses: Helps deliver sanitized and more meaningful AI-generated responses, further validating Maven compilations and dependency resolutions.
• Varying levels of AI-assistance: Users can choose which issues gen AI solves based on the estimated effort an issue will require.

For additional help, Konveyor is also introducing various updates to help with replatforming applications to Kubernetes. For example, a new asset generation feature allows for application deployment and runtime configuration retrieval and then creates Kubernetes deployment artifacts. 

Open source and AI: A hybrid approach

For over 30 years, open source technologies have fueled rapid innovation while reducing IT costs. Red Hat’s contributing engineers continue to drive this vision by bringing generative AI to the Konveyor project, helping to change the economics of replatforming and refactoring applications for the open hybrid cloud. 

With these advancements, the Konveyor project is showcasing how organizations can unlock new modernization efficiencies while maintaining required levels of flexibility and compliance, as well as a strong security footprint. The hybrid cloud is the foundation for modern applications, and AI thrives on the open hybrid cloud. 

These new Konveyor generative AI capabilities are available now. Learn more or start using immediately. Additionally, Konveyor’s asset generation feature will be readily available this summer and Red Hat plans to enhance its  migration toolkit for applications in future releases.