Brainstormer — Interactive Whiteboard with Multi-Modal AI Assistant

Brainstormer is an innovative tool designed to enhance your brainstorming sessions, system design, and creative thinking by combining a dynamic, interactive whiteboard with the power of AI. Whether you're sketching out complex software architectures or generating new ideas, Brainstormer seamlessly integrates a smart whiteboard with AI-driven insights to make your process more efficient and visually intuitive.

• Interactive Whiteboard: Powered by Excalidraw, Brainstormer offers a flexible and easy-to-use whiteboard for sketching ideas, system designs, and more.
• AI-Driven Chat: Integrated with an LLM like Ollama, Brainstormer provides intelligent responses and suggestions based on your whiteboard sketches and text inputs.
• Multimodal Interaction: Export your whiteboard as an image and use it as context within the chat, allowing the AI to provide more relevant and targeted advice.
• Responsive Design: The layout adapts to different screen sizes, providing a seamless experience whether on desktop or mobile devices.

• Node.js (v14 or higher)
• npm (v6 or higher)

1. Clone the repository:

   git clone https://github.com/shivangdoshi07/brainstormer.git
   cd brainstormer

2. Install the dependencies:

   npm install

3. Start the backend server:

   cd backend
   node index.js

4. Start the frontend development server:

   cd frontend
   npm start

5. Open your browser and navigate to http://localhost:3000 to start using Brainstormer.

Brainstormer can be deployed using Docker or Podman with the provided docker-compose.yml file.

• Docker or Podman installed
• Docker Compose or Podman Compose

# Build and start containers
docker-compose up -d
# or with Podman:
podman-compose up -d

# Access the application at http://localhost

Create a .env file based on .env.example:

cp .env.example .env

Edit the .env file to configure your LLM provider and models.

• Frontend: http://localhost (port 80)
• Backend API: http://localhost:5000

• Set System Prompt Using User Input: Allow users to customize the system prompt for the AI, tailoring responses to specific needs and preferences.
• Pass Chat History to the Chat Completion API: Instead of only sending the most recent message, the full chat history will be passed to the API, enabling the AI to maintain context over longer conversations.
• Convert Whiteboard Elements to Text: Enhance the AI's understanding by exporting scene elements from Excalidraw and converting them to text, rather than just using an image.
• Add LLM Router: Implement a router that intelligently directs chat input to either a pure LLM or a vision-enabled LLM based on the type of question.
• Function Calling for Whiteboard Updates: Introduce function calling capabilities, allowing the AI to directly add new elements to the whiteboard based on the conversation.

• Excalidraw for the whiteboard functionality.
• Ollama for the LLM API integration.