Most large enterprises are burdened with sprawling legacy application environments that need to be modernized for the public cloud. Many have already begun the journey. Yet, according to Vladimir Agres, Vice President of Cloud Business at EPAM, too many modernization initiatives stall, fail to deliver the expected outcomes—or are abandoned entirely.
“Most migration and modernization projects stop or stall because organizations underestimate the complexity and misalign the work with business outcomes,” says Agres. “What was assumed to be a simple lift-and-shift quickly becomes application replatforming, and that in turn snowballs into complex, cloud-native modernization.”
The challenge, Agres notes, is compounded by the nature of legacy systems: millions of lines of code, often written in outdated versions of multiple programming languages and running on obsolete infrastructure. In many cases, organizations have substantial enterprise agreements with cloud providers but are unable to take full advantage of these investments, as their development teams are often challenged by a lack of skills and capabilities, clear documentation, and sound delivery frameworks.
The rise of gen AI development tools offers a potential breakthrough—but Agres cautions that “out-of-the-box” AI coding agents aren’t a silver bullet. There are challenges: out-of-the-box agents and vibe coding are not yet sufficient for legacy modernization. “Vanilla AI tools lack the context required for complex legacy modernization,” Agres explains. “The most dramatic productivity gains come from custom-built AI agents and orchestrated multi-agent workflows—but most organizations can’t implement these alone. And some veteran engineers are still skeptical, which makes change management essential.”
Agres likens it to the long road to adopting automated testing: “It took years for developers to trust automated tests. Now we’re asking them to trust AI tools that generate code.”
This is where EPAM leverages AI/RunTM, a framework covering the end-to-end product development life cycle, that addresses these challenges. Designed to embed AI across the entire product development lifecycle (PDLC), AI/RunTM helps organizations accelerate application modernization efforts. “We don’t focus on tools but rather on adoption, acceptance, and transparency which, in turn, builds trust and measurable business value,” he says.
With the AI/RunTM Framework, organizations can:
- Build custom AI agents tailored to their systems and business goals.
- Orchestrate multi-agent workflows to accelerate and derisk modernization.
- Integrate seamlessly with best-of-breed AI tools like AWS Bedrock, Amazon Q Developer, and AWS Transform.
- Govern and measure adoption through AI literacy programs, ROI tracking, and adoption dashboards—ensuring developer trust and organizational alignment.
Case in point: A large healthcare provider recently turned to EPAM’s AI/RunTM to modernize a monolithic medical imaging platform containing millions of lines of code. The goal involved transforming the system into a cloud-native SaaS application. The challenge: decades-old Java, .NET, and Perl code tightly coupled with other legacy databases and tools.
Using AI/RunTM, EPAM deployed custom AI agents that:
- Mapped application dependencies by scanning hundreds of thousands of connections and identifying safe disconnect points.
- Converted extensive Perl scripts into Python in just two weeks—a task that would have taken months to complete manually.
- Leveraged AWS Bedrock and Amazon Q Developer for context-aware code transformation and refactoring.
Within weeks, the client saw a significant increase in modernization velocity, and—crucially—internal developers began embracing the AI-powered workflows rather than resisting them.
While many organizations are still experimenting with standalone AI tools, EPAM’s AI/RunTM offers a cohesive, end-to-end approach that unites the right best of breed technology with the right methodology.
“We don’t just hand our clients tools—we help them orchestrate them across the entire PDLC,” says Agres. “That’s the power of the AI/RunTM Framework.”
By combining custom AI engineering, process orchestration, and developer adoption strategies, AI/RunTM not only accelerates legacy modernization but also builds a foundation for future AI-native development.