RAG Chatbot Hackathon: Internal Wiki
One-day hackathon for 20 developers to build a RAG chatbot on internal Wiki using AI coding tools
The Challenge
Organizations need to make internal knowledge instantly accessible. Manual search through Wiki documentation is slow and inconsistent. This one-day hackathon brought together 20 developers—backend, frontend, and test engineers across different languages—who had no prior AI knowledge and did not code with AI on a daily basis. The programme was tailored to their experience so they could design and implement a RAG-powered chatbot that answers company questions from Wiki docs, cites sources, and says when it doesn't know.
Our Approach
The hackathon opened with a theoretical introduction to RAG: how it works, what a vector datastore is, and the end-to-end flow from documents to cited answers. This gave developers the conceptual foundation before any coding.
Coding with AI was built into the programme. Teams produced both backend and frontend using AI-assisted tools such as Cursor and GitHub Copilot, learning effective prompting and iteration as they went.
Developers then designed and implemented their RAG chatbot. It had to answer simple and complicated questions from the internal Wiki. When the answer was not available in the docs, the system did not hallucinate—it explicitly stated that it didn't know. A key requirement was to list the relevant contact person and attach links to the source Wiki documentation.
The full pipeline—load docs, chunk, embed, vector store, retrieve, LLM, and a simple chat UI—was completed in one day. Teams presented their architecture, demo, and learnings on functionality, UX, and how they used AI tools to build faster.
The Results
Teams learned how to quickly build prototypes with AI coding tools and how to build RAG chatbots and how they work. Developers without prior AI experience left with a working chatbot that answers from Wiki docs, cites sources and contacts, and refuses to hallucinate when the answer isn't in the knowledge base.
20 developers (backend, frontend, tests) Attendees
1 full day Duration
Prototypes with AI tools + RAG know-how Outcome
Technologies Used
Ready to build something similar?
Let's discuss how we can apply these approaches to your specific challenges.
Get in Touch