One of the world’s leading automotive services companies faced a paradox: whenever engineers needed to modify a feature or develop new software, they spent 40–60% of their time simply searching for information across a massive codebase. FPT turned to AI to help its U.S.-based partner tackle this complex challenge.
The company, one of the world’s largest automotive services and technology corporations headquartered in the United States, operates software systems containing millions of lines of code accumulated over multiple generations. Written partly in legacy programming languages and repeatedly extended over the years, the system had become increasingly complex.
It was like navigating a vast underground city without a complete map: even a minor change could trigger a chain reaction across the system. Engineers therefore had to spend significant time tracing complex dependencies, assessing potential risks, and developing a deep understanding of the system before making changes safely.
The pressure has only intensified as AI continues to shorten technology innovation cycles. Smaller companies can experiment and iterate quickly, while large corporations must proceed more cautiously because every change can directly affect critical operations.
To support engineering teams, many enterprises have adopted AI-powered Q&A tools and coding assistants. Initially, these tools helped engineers retrieve information and complete tasks more quickly. But as enterprise data grew larger and more fragmented, their limitations became increasingly apparent.
Without a comprehensive understanding of the entire system context, AI could generate incomplete or inaccurate analyses—or even produce plausible-sounding answers that were factually incorrect. At the same time, operating costs could rise rapidly as the AI had to process enormous volumes of data.
One of FPT’s approaches to addressing this challenge is Flezi Metis, an AI platform that connects source code, technical documentation, and operational data into a unified knowledge map of the entire system.
With a holistic view of the technology landscape, AI can reason based on the system’s actual architecture and relationships rather than making assumptions. This helps reduce inaccurate outputs caused by missing context while eliminating the need to repeatedly process the entire system whenever a new request arises.
Flezi Metis was developed by FPT.AI1, an FPT unit specializing in AI solutions. The platform applies AI to transform how enterprises understand, operate, and modernize large-scale, complex IT systems.
When an engineer needs to modify a function, Flezi Metis can quickly identify related components, analyze potential impacts, and flag areas that may be affected by the change. Customer data remains within the customer’s own infrastructure, helping meet stringent security requirements.
In an initial pilot with the U.S. automotive customer on a project containing approximately one million lines of code, Flezi Metis delivered measurable results: source code search time was reduced by approximately 50%, while overall task completion time for the engineering team decreased by around 30%.
Beyond helping engineers work faster, Flezi Metis also changes how enterprises manage complex software systems. When institutional knowledge becomes easier to access, technology leaders gain a more comprehensive view of their systems and can make better-informed decisions.
Ms. Nguyen Thi Quynh Lan, Director of the AI Solutions and Business Center, the unit behind Flezi Metis, together with her colleagues at FPT Software.This challenge is far from unique to one company. After years of working with customers across the United States, Europe, and Japan, FPT has identified it as a common issue among large enterprises, where critical operational knowledge—from business logic to troubleshooting experience—can become buried deep within legacy codebases or remain largely dependent on the knowledge and experience of veteran engineers.
In Japan alone, IDC estimates that approximately 80% of medium-sized and large enterprises still rely on legacy systems.
Following its successful deployment with the U.S. customer, Flezi Metis is now expanding into sectors with complex operational requirements, including banking, insurance, and the public sector.
The platform forms part of FPT’s FleziPT AI ecosystem, which brings together technology platforms, experts, and partners to enable enterprise-wide AI transformation.
To further accelerate AI transformation, FPT has also introduced CASAN, a five-level AI maturity framework designed to help enterprises standardize their transformation roadmaps—from fragmented, standalone AI applications to operating models where AI is deeply integrated across the entire organization.
As FPT moves into 2026, every line of code and every "Made-by-FPT" solution goes beyond solving individual business challenges. They also carry a broader mission of “co-creating national capabilities”, harnessing technological intelligence to strengthen Vietnam’s digital self-reliance.