Momentum Group needed to bring together complex financial and entity data from across approximately 300 entities to support its Organisation for Economic Co-operation and Development (OECD) Pillar Two tax compliance requirements. Entelect used Claude Code to rapidly prototype a custom data mapping tool and build AI-driven pipelines that analyse unfamiliar data sources and generate the code needed to onboard them. The initial tool was prototyped in approximately two weeks rather than several months, while new data sources can now be onboarded in roughly one to two hours instead of days of manual development.

With Claude, Momentum and Entelect:

• Prototyped the initial Pillar Two compliance data pipeline in approximately two weeks instead of several months
• Reduced new data source onboarding from days of manual development to roughly one to two hours
• Onboarded seven data sources to date
• Used Claude to analyse unfamiliar data sources and automatically generate the fields and company breakdown needed to onboard them
• Used Claude to generate the reader and mapper code required to import new sourcesUsed Sonnet for day-to-day prototyping and pipeline execution, with Opus supporting the initial design and generation of the pipelines

Entelect developed a custom data mapping tool that consolidates Momentum's different source data into a consistent structure for the Pillar Two compliance process. Claude Code was incorporated into the engineering workflow from the beginning, helping rapidly prototype the tool's command-line interface, application programming interface layer and Tauri/React user interface.

Claude was then used to address one of the more repeatable challenges in the process: onboarding a new and unfamiliar data source. A Claude-driven source-analysis pipeline inspects the source and automatically generates the fields and company breakdown required to understand and onboard it. A second mapper-generation pipeline uses that analysis to generate the reader and mapper code needed to import the source into the consistent data structure.

This means that each new source does not require the same level of manual development from scratch. Claude handles much of the initial source analysis and mapper generation, while the resulting pipeline provides a repeatable way to extend the tool as new data sources are introduced.

The team used Claude Code through the Visual Studio Code extension, with Sonnet handling most of the prototyping and day-to-day execution of the source-analysis and mapper-generation pipelines. Opus was used for the initial design and generation of those pipelines, allowing the team to use different models according to the needs of the task.

The initial Pillar Two compliance data pipeline was prototyped in approximately two weeks, compared with the several months estimated for a conventional approach. More importantly for the ongoing compliance process, onboarding a new data source has fallen from days of manual development to roughly one to two hours using the Claude-generated analysis and mapper pipelines.

Momentum has onboarded seven data sources to date, with the same Claude-driven process available as additional sources need to be incorporated. Instead of repeatedly analysing each source and manually developing the corresponding mapping logic, the team can use the established pipelines to accelerate much of that work.

Beyond the development efficiencies, the project helped Momentum's Group Tax team establish a more structured approach to bringing together the information required for Pillar Two compliance. The mapping tool provides a consistent path for consolidating disparate data into the required format, while Claude makes that capability easier to extend as the underlying data landscape evolves.