Legacy systems have long been a bottleneck for AI-assisted development — until now, the missing piece was runtime context. vFunction’s latest release aims to close that gap.
The cost of waiting to modernize these applications is going up fast, and the cost of modernizing has never been lower.”— Moti Rafalin, vFunction CEO & Co-Founder
vFunction has unveiled vFunction 5.0, a new version designed to pull large-scale Java and .NET applications into the AI-driven development fold.
This update introduces specifications and integration tests that are automatically generated from the running application. It also streamlines the migration of SQL Server, Oracle, and DB2 databases to PostgreSQL within the same workflow. These outputs are then delivered to AI coding platforms such as Kiro, Claude Code, and AWS Transform through vFunction’s MCP server and downloadable skills.
Also worth a look: AiRET’s AX1 shoe care appliance decides how to clean each pair on its own and NeuPT’s ReadyOS beta clinics show readiness data can transform PT visits.
AI coding tools are strong at generating code, but they struggle with large existing applications due to a lack of understanding of the system’s current behavior, no agreed-upon target architecture, and no method to verify that changes haven’t caused regressions. vFunction 5.0 tackles all three by observing production behavior, enabling architects to define a target state with supporting evidence, and producing specs and tests from that observed behavior. The platform is specifically designed for applications exceeding roughly 500,000 lines of code or 1,000 classes.
“The cost of waiting to modernize these applications is going up fast, and the cost of modernizing has never been lower. With 5.0, a team can modernize its largest application domain by domain and verify each change before the next one begins,” said Moti Rafalin, CEO and co-founder of vFunction.
Key additions in vFunction 5.0
– Specifications derived from production behavior. Each domain within an application receives a specification based on analysis of the live system. Every element is backed by runtime evidence or code-derived data, and architectural issues flagged for that domain are incorporated, so newly generated code avoids repeating them.
– Automated integration test generation. Tests built from actual production traffic provide substantial integration coverage before any code is changed, even for applications that previously had no tests, ensuring new code matches old behavior.
– Database migration to PostgreSQL. SQL Server, Oracle, and DB2 databases can now be migrated as part of the same workflow: schema and SQL dialect conversion, driver and connection pool updates, ORM switching to Hibernate for Java or EF Core for .NET, and AI-driven conversion of stored procedures into application code. Data movement itself remains handled by the customer’s existing database tools.
The release includes a downloadable skill that generates code from a specification one domain at a time, along with direct installation and authentication for Claude Code and Kiro command-line tools. AWS Transform for Java workflows is now supported alongside .NET, and a new Modernization tab enables the end-to-end workflow in a single interface. More details are available on the vFunction website.
“A specification inferred from code repeats the code’s problems. In 5.0 the specification and the tests come from the runtime behavior, and take into account observed architecture issues, which lets a coding tool modernize a domain of a half-million-line application while reducing its dependencies on other modules,” said Amir Rapson, CTO and co-founder of vFunction.
vFunction 5.0 is delivered in the same manner as every vFunction engagement: vFunction’s forward-deployed engineers run the process alongside the customer’s team, inside the customer’s environment, until the team can manage it independently. The combination of vFunction’s modernization harness, top-tier AI coding tools, and vFunction’s FDEs enables rapid, low-risk, fixed-cost modernization initiatives.
Availability
vFunction 5.0 is now available to existing customers. The platform installs within the customer’s own environment — on-premises or in their cloud tenancy — and can be accessed via AWS Marketplace, Microsoft Azure Marketplace, and Google Cloud Marketplace.
About vFunction
vFunction helps enterprises modernize existing applications with AI. It provides AI coding tools with the context and guardrails needed to refactor, rewrite, and modernize existing systems with speed, accuracy, and control. By pairing AI coding tools with runtime architectural context and proprietary data science, vFunction supports key modernization use cases, including full-stack Windows modernization, large VM-to-container migrations, and AI-driven development for critical applications. Learn more at vfunction.com.
Paul Gordon
vFunction
Visit us on social media:
LinkedIn
—
**Why it matters:** As AI coding assistants become more powerful, their usefulness on legacy systems has been limited by a lack of runtime insight. vFunction 5.0 demonstrates a path forward: by grounding AI tools in observed production behavior, enterprises can modernize large, mission-critical applications with lower risk and greater speed — potentially reshaping how legacy modernization is approached across industries.
