Chatbot

ChitChat with Chehak – Your smart, secure AI companion for meaningful conversations. Built with intelligence, designed for privacy.

Comment

Inspiration

The idea for ChitChat with Chehak came from a simple thought:

“What if anyone could have a personal AI assistant that felt human, helped intelligently, and respected their privacy?”

With the rapid rise of conversational AI, I wanted to build something more than just a chatbot — a secure, multimodal AI companion that could understand text, images, documents, and even voice. Inspired by tools like ChatGPT and personal experiences with academic and project queries, I aimed to create a localized, privacy-focused AI tool that users can fully trust.

What it does

ChitChat with Chehak is a full-featured local AI chatbot that allows users to:

  1. Chat using text or speak via voice input
  2. Upload documents (PDF, DOCX, TXT) and ask questions based on the content
  3. Upload images and get AI-generated insights or explanations
  4. View and maintain chat history locally on the device
  5. Use the entire system without internet, as it runs on local models (via Ollama)
  6. Get fast, intelligent answers with complete data privacy and security

The app acts as a personal AI assistant that understands multiple input types and responds intelligently, while keeping everything user-side.

How we built it

Tech Stack Frontend: HTML, CSS, JavaScript Backend: Node.js + Express.js AI Engine: Ollama with local LLaVA / LLama3 model File Handling: Multer, pdf-parse, mammoth Voice Input: Web Speech API Security: Local storage, no cloud/API data sharing

Challenges we ran into

Every good project comes with its hurdles, and here’s what I tackled:

  1. Streaming large responses from local models while keeping the UI responsive
  2. Managing stateful conversation history without using a database
  3. Parsing complex document formats and returning meaningful responses
  4. Handling CORS issues and file size limits in backend routing
  5. Implementing cross-browser-compatible voice input features

Each obstacle sharpened my problem-solving skills and pushed me closer to production-grade development practices.

Accomplishments that we're proud of

1,. Created a fully functional, multimodal chatbot without using any paid APIs or cloud services

  1. Successfully integrated document and image understanding with local AI models
  2. Built a voice-enabled assistant that feels interactive and human
  3. Ensured zero data leakage by handling everything locally on the user’s device
  4. Designed a smooth, responsive chat interface that supports streaming answers and input from multiple sources
  5. Overcame real-world challenges like CORS errors, large file handling, voice compatibility, and streaming performance

What we learned

Throughout the development of this project, I deepened my understanding of:

  1. Frontend-backend integration using JavaScript, HTML, and Express.js
  2. Working with Ollama and LLaVA/LLama3 APIs for free, local AI responses
  3. Handling file uploads (PDF, DOCX, TXT, Images) and parsing them dynamically
  4. Integrating voice input features using the Web Speech API
  5. Efficiently managing chat history, streaming responses, and state updates
  6. Implementing data security measures to keep all user data local and private

What's next for Chatbot

ChitChat with Chehak is just the beginning. I aim to:

  1. Add multilingual support for global accessibility
  2. Enable chat export and session sharing
  3. Explore offline Progressive Web App (PWA) capabilities
  4. Continue refining AI response accuracy
Share this project:

Updates