The problem I'm solving is that moving customer information into new systems when setting up accounts is usually slow and full of mistakes, which frustrates everyone involved.<p>I happened to have a research project on general data transformation.
So I built a quick demo to tackle this issue.<p>The demo uses Claude 3.5 and a prompt framework to transform customer CSV files into your application's data model. The UI is built with Streamlit.<p>I started with CRM data as an example, but I'd love to expand to more complex use cases with various data types. If you have interesting scenarios, let me know.<p>Here's the research behind it:
<a href="https://cocoon-data-transformation.github.io/page/transform" rel="nofollow">https://cocoon-data-transformation.github.io/page/transform</a>
The project is open sourced:
<a href="https://github.com/Cocoon-Data-Transformation/cocoon">https://github.com/Cocoon-Data-Transformation/cocoon</a>