Answer, RAG visualization on documents

The project

Answer is a fork of Dataiku’s Answer project, which ships a turnkey RAG visualization web app. Day to day, a fair amount of features were missing, and my clients kept asking for improvements. So I took matters into my own hands and forked it to plug the gaps. Typically the kind of situation where you could wait for an official release, but in consulting, that’s not really how it works.

My contributions

Forked and extended the original app: added client-requested features, improved the RAG visualization, fixed issues and adapted it to the real-world use cases we hit in enterprise settings. The idea was to keep a version that’s usable right away, without waiting for upstream to move.

Takeaways

Taking an existing open source project and adapting it to real field needs is a great exercise. It helped me understand RAG in depth, and what enterprise users actually want, which is often pretty far from the demo cases you see in presentations.

Context

Professional project. As a consultant I kept noticing the same gaps in Dataiku’s RAG visualization tool. Rather than wait, I forked it to address them, even if it means re-syncing later when upstream catches up.

Tech stack

• Python

• Node.js

• Gulp (build system)

• Playwright (tests)

• OpenAPI

• Dataiku

Cover image generated by Nano Banana (Google), since I no longer have the mockup template I used before. So it’s not a screenshot of the actual tool.