Key Outcomes
Up to 60x faster than manual migration
Complex legacy pipelines automatically translated, documented, and cloud-ready
Data owners and data engineers can take strategic decisions thanks to information and insights gained from the AI tool
Lower costs and a team built to scale independently
The story in a glance
Unieuro, Italy's leading consumer electronics retailer, partnered with Thoughtworks and Huware to modernize its legacy data infrastructure ahead of a major European expansion.
Faced with hundreds of undocumented data pipelines and the complexity of migrating to Google Cloud Platform, the team built an AI-powered accelerator that automated both code translation and documentation generation. This reduced a migration effort that would have taken years manually to just a few months, leaving Unieuro with a scalable platform built for the future.
About the client
Unieuro S.p.A. is Italy’s leading retailer of consumer electronics and household appliances. Founded in 1967 as a small warehouse business in Alba, the company has grown with a vast network of over 500 physical stores and, with a growing e-commerce presence, generates €2.6 billion in annual revenues. A majority acquisition by Fnac Darty marked the next chapter of Unieuro’s growth, with ambitions to scale the company across Southern and Western Europe. To support this endeavor, Unieuro needed to modernize the data structure underpinning its operations, a significant data modernization challenge. They turned to technology partner Huware and the team at Thoughtworks to make it happen.
The challenge
How can a legacy system be migrated without any supporting documentation?
Unieuro’s data operations had grown organically over decades, built on on-premises integration tools that were never designed to operate at the scale the business now found itself at. The data pipelines at the heart of the operations were complex and deeply embedded with business logic and used both for data integration and data analysis.
Moving to Google Cloud Platform was the clear strategic direction, providing the opportunity to rebuild the data platform while addressing data migration challenges. But without a way to automatically interpret and translate the legacy pipelines, the only option would have been to reverse-engineer each one by hand — a slow, high-risk process entirely dependent on expertise that was stretched thin. With hundreds of pipelines to migrate to Google Cloud Platform, that approach simply wasn't viable. Unieuro partnered with Thoughtworks and Huware, leveraging AI to enable a smarter path forward.
The solution
How to build a scalable AI-powered engine for data pipeline migration
Thoughtworks, working in close partnership with technology partner Huware, developed an MVP built around two core engines in only three months: an automated pipeline migration and self-generating documentation.
Rather than migrating each pipeline manually, the team built a system powered by AI automation that could do the heavy lifting autonomously. Data owners and data engineers used the tools to automate the data pipeline migration.
Thoughtworks and Huware brought their co-delivery expertise in AI engineering, data platform management, and a structured delivery approach. The two teams worked in tandem throughout the process, with Thoughtworks leading the initial delivery while upskilling the Huware and Unieuro teams to own and extend the digital solution independently.
"What stood out about our partner Thoughtworks was how quickly they got to grips with the complexity of the problem. From day one, they were focused on building something that would last beyond the engagement, not just delivering to the brief. "
Andrea Ciccione, Data & Cloud Unit Manager, Huware
The solution was structured around two parallel streams. The first was an AI-powered translation layer, using Gemini models to convert its ETL legacy system jobs into SQL for BigQuery and Dataform. A custom code parser handled the proprietary ETL legacy system XML format, and few-shot learning and dynamic prompting techniques ensured the output was accurate and contextually aware. We are working to automate the validation, running source-to-target comparisons to verify the translated data pipelines against the originals.
The second stream tackled the documentation gap head-on. Every pipeline migration automatically triggered the generation of both technical and business documentation, published directly to Confluence. This was transformative, with knowledge finally captured, standardized, and accessible.
Together, these two streams created a reusable accelerator that Huware could use across Unieuro’s broader legacy system, making future migrations significantly faster and lower risk.
"The collaboration worked because both teams were focused on the same outcome. We came in with strong domain knowledge, and Thoughtworks brought the AI-first delivery framework based on their international expertise. The combination is what made it work."
Andrea Servili, CEO & Founder, Huware
Key outcomes
Up to 60x migration acceleration compared to manual data pipeline migration.
Automated the full process from code translation to documentation generation.
Built lasting capability with lower costs, greater predictability, and no single points of knowledge.
Leaving it to Data Owners and Data Engineers to decide the migration strategy to apply for specific data pipelines.
The outcome
How Unieuro cut migration time and built a foundation for the future
The impact of Huware and Thoughtworks’ project was immediate. By automating what had been a slow process that could only be undertaken by people with a deep understanding of Unieuro’s legacy systems, the new solution removed a major delivery bottleneck. It delivered migration speeds up to 60 times faster than the manual process.
Beyond raw speed, the project delivered lasting structural benefits including parallel migration capability, standardized documentation practices, and lower legacy licensing and maintenance costs. Now, Unieuro can deliver data and migrate pipelines at scale. It avoided a resource-heavy lift-and-shift approach, and Huware’s team is well equipped to continue the work based on business needs in the future.
We knew migration was going to be complex and expected to strategically adopt AI to facilitate it, but we didn't expect how quickly the team got up to speed and started delivering. Having Thoughtworks and Huware working together meant we had both the technical depth and the hands-on partnership to make it work.
Thanks to the flexibility of the AI toolkit we developed, we are able to operate both in a Lift & Shift approach and in a legacy ETL modernization mode—transitioning toward ELT in a cloud-oriented context. This is also enabled by integrating the tool within the data delivery life-cycle and decision-making processes across data teams.
Looking ahead
How data modernization can lead to enterprise-wide scaling
The translation pipeline built by Thoughtworks offers a proven template for tackling legacy migration and complexity at speed. The three-month engagement between the two organizations was designed as a launchpad for Unieuro’s systems, not an endpoint. The ability to move quickly, share data intelligently, and operate on modern Google Cloud Platform infrastructure will be a strategic differentiator and give Unieuro the capacity to expand further than ever before, for years to come.
