Minutes, not months: how Wynxx accelerates legacy modernisation in financial services
95% faster documentation creation
€300K in value delivered in one day
500% ROI in the first month
Challenge
A new approach to an old challenge
In financial services, legacy systems are both an asset and a burden. Built over decades to meet growing demands, these applications underpin mission-critical processes – yet their complexity can make them difficult to understand, maintain, and evolve.
For one leading global bank, this challenge had become particularly acute. Senior engineers were spending up to four weeks manually producing documentation for a single legacy application, analysing thousands of lines of code line by line. Modernisation projects slowed, compliance reviews became time-consuming, and valuable expertise risked being lost as experienced staff moved on.
The bank needed a faster, more reliable way to gain insight into its systems – one that improved accuracy and efficiency without compromising trust or security.
About the client
The client is one of the world’s largest financial services organizations, serving millions of customers across retail, corporate, and investment banking. With a presence in over 50 countries, the bank is recognized for its commitment to innovation, digital transformation, and responsible finance. Its strategic focus lies in strengthening its digital ecosystem to deliver secure, efficient, and client-centric financial solutions globally.
Engagement
Wynxx – Generative AI for the Software Development Life Cycle
Lorem ipsum
GFT Technologies introduced Wynxx, the industry’s most flexible generative AI platform for enterprise software delivery.
Wynxx accelerates enterprise software transformation by making generative AI operationally viable across the Software Development Life Cycle (SDLC) – responsible AI by design, model-agnostic, and measured by business outcomes. It integrates seamlessly with enterprise tools and infrastructure, enabling teams to focus on high-value work while accelerating delivery and improving quality.
To demonstrate its value in a controlled and measurable way, the team adopted a three-sprint approach. This agile framework enabled them to establish a clear baseline, test Wynxx on real applications, and assess performance and cost objectively. Each sprint built on the insights of the previous one to showcase scalability and accuracy at enterprise level:
Defined metrics and prepared the environment
Defined metrics and prepared the environment
Tested Wynxx on a “lighthouse” legacy application
Tested Wynxx on a “lighthouse” legacy application
Measured quality, efficiency, and business impact
Measured quality, efficiency, and business impact
Engagement
From manual to automated
The lighthouse application selected was a 20-year-old Java system – complex, with multiple upstream and downstream connections.
Using Google Cloud Vertex AI and Gemini 1.5 Pro, Wynxx ingested the source code directly from the bank’s repository and generated structured outputs aligned with the bank’s standards, including:
-
Architecture and data-flow diagrams
-
Component-level explanations
-
Natural-language functional summaries
What had previously taken weeks was now completed in minutes. The generated documentation was iteratively reviewed and refined through prompt engineering and expert feedback until it met – and in several areas exceeded – the quality benchmarks defined at the start of the project.
Benefit
Measured, verified results
Lorem ipsum
After validating the initial use case, the team expanded the experiment to ten additional legacy applications, demonstrating Wynxx’s scalability and consistency.
Overall impact:
-
95% time reduction in documentation time
-
93% lower unit cost per artifact
-
500% ROI within the first month
-
€300,000 in equivalent value delivered in a single day
The enormous cost optimisation and efficiency gains achieved through Wynxx highlight what’s possible when generative AI is applied responsibly in complex environments.
These results not only demonstrated significant efficiency gains but also validated the accuracy and reliability of Wynxx’s AI-generated outputs.
Accuracy averaged 82%, with some outputs reaching 100%. Completeness averaged 71%, reflecting the natural limits of code-only analysis. Automatically generated sequence diagrams showed process flows and data exchanges more precisely than manually created versions.
The cost of AI execution was negligible – approximately $0.50 per API call – with human validation retained for quality assurance.
Human and operational impact
The efficiencies achieved went far beyond time saved. By removing repetitive documentation work, engineers could focus on development and innovation. Job satisfaction rose, compliance teams reduced audit-prep effort by up to 80%, and technical knowledge became transparent and shareable across teams.
Beyond documentation: scaling AI across the SDLC
Following the pilot, Wynxx became a blueprint for enterprise-wide modernisation. Its model-agnostic foundation preserves choice and control, allowing integration with existing enterprise toolchains to support:
- Code Quality Analysis: Identifying vulnerabilities and improvement opportunities.
- Automated Test Generation: Accelerating regression and functional testing.
- Architecture Recommendations: Guiding modernization pathways.
- Predictive Maintenance: Detecting anomalies before failures occur.
Wynxx has since evolved from a documentation accelerator into an AI co-pilot for continuous software engineering improvement – accelerating delivery, reducing cost, and ensuring consistency across the SDLC.
Modernisation at scale
For financial institutions operating under tight regulatory and competitive pressure, Wynxx demonstrates that generative AI can deliver measurable, trustworthy results today.
By automating and improving time-intensive development tasks, Wynxx helps organizations understand, maintain, and modernise their systems faster – transforming legacy complexity into a foundation for innovation.
