Internal RAG Assistant
An enterprise-grade GenAI tool that reduced technical report writing time from 2 weeks to minutes. Features a live editor with precise source citations.
Year
2025
Stack
The Challenge
Engineers were spending 1 to 2 weeks manually searching through thousands of legacy PDF documents to compile a single 40-page technical report. The process was slow, error-prone, and inefficient, tying up valuable resources on administrative tasks rather than engineering.
The Solution
We developed a secure, internal Retrieval-Augmented Generation (RAG) application. The goal was not just to "chat" with documents, but to produce a professional, editable deliverable.
- Smart Editor: The LLM generates text directly into a rich-text editor, allowing engineers to refine and format the report instantly.
- Traceability: Crucially, every generated section includes page-level citations. Users can hover over a sentence and see exactly which PDF and page it came from, ensuring 100% accuracy and trust.
- State Management: Utilized complex Redux logic to handle chat history, streaming responses, and large datasets of citations without compromising UI performance.
The Impact
The tool dramatically accelerated the workflow, allowing engineers to generate a draft structure in minutes instead of days, while maintaining strict security via SSO and RBAC protocols.