QuickOnboardDoc
An AI onboarding assistant that answers staff questions from a company's own documents using RAG, with multi-tenant workspaces and Google sign-in.
Fullstack Development
1 month
Introduction
An AI onboarding assistant that answers staff questions from a company's own documents using RAG, with multi-tenant workspaces and Google sign-in.
Work Done
Full System Build
Challenges
Creating a highly accurate AI system that would only answer questions based on the uploaded documents and refuse to guess, solving the common problem of AI 'hallucination.'
Solutions
I implemented a RAG (Retrieval Augmented Generation) system using LangChain and Gemini to ensure all answers are directly sourced from the documents. This delivers a confidence level and the source file for every accurate answer.
Results
A fully functional, secure web application that can be installed as a PWA. It successfully provides accurate, instant answers to document questions, offering a centralized solution to a common HR and development problem.
Details
Answering from the docs, not from thin air
QuickOnboardDoc is an AI onboarding assistant that fixes information overload. Instead of making new hires read endless PDFs, they ask the assistant, and it answers from the company's actual documentation.
RAG, so it doesn't guess
This isn't a plain chatbot. It's a RAG system that reads the uploaded company documents. Before answering, it retrieves the relevant context from the files (stored in Vercel Blob), cites the source document it used, and is built to say when it doesn't have the answer rather than invent one. For questions about company policy, that restraint is the whole point.
Built for real organisations
The app is multi-tenant. A user can manage several companies as separate workspaces, each with its own data and documents, fully isolated from the others. Sign-in runs through NextAuth with Google OAuth, so only authorised staff reach the proprietary data.
The experience
I gave it a dark-mode-first look (I called the theme "Crypt Keeper") for a developer-centric feel, with real-time typing indicators and responses that come back in about a second.
Video Showcase
Testimonials
This bot solves a huge problem! Being able to upload a PDF and get instant, reliable answers without having to bother a senior developer is a massive time-saver.
Tope Akintola
November 28, 2025
Fullstack Development
