Botly

Botly is a SaaS platform for creating, managing, and deploying AI-powered chatbots that you can easily embed in your website or React app. It provides a user-friendly dashboard, robust backend, and a customizable React component (<BotlyBot />) available as an npm package.

---

Retrieval-Augmented Generation (RAG) for Dynamic LLM Prompting

Botly leverages a Retrieval-Augmented Generation (RAG) architecture to enhance the contextual relevance and factual accuracy of LLM responses. Upon receiving a user query, the system dynamically retrieves pertinent context vectors from a vectorized knowledge base (e.g., website embeddings, CSV data, or uploaded documents) using semantic search algorithms (such as cosine similarity over embeddings generated by transformer-based models). The retrieved context is then injected into the prompt as system or context messages, enabling the LLM (e.g., Gemini, Grok, DeepSeek) to ground its generation in up-to-date, domain-specific information. This hybrid approach mitigates hallucinations, supports long-tail queries, and enables scalable, domain-adaptive conversational AI.

Key components:

• Vector Store: Contextual data is embedded and indexed for efficient similarity search.
• Retriever: On each user input, the retriever fetches top-k relevant context snippets.
• Prompt Composer: Dynamically constructs LLM prompts by concatenating retrieved context with user queries.
• LLM Integration: Supports multiple LLM backends, with context window management and fallback strategies.

This RAG pipeline ensures that Botly delivers accurate, context-aware, and up-to-date responses tailored to each user interaction.

Features

• No-code Bot Creation: Build chatbots with custom context, capabilities, and personality using a simple form.
• Embed Anywhere: Use the botly-bot npm package to add your chatbot to any React app.
• AI-Powered: Supports multiple LLMs (Gemini, Grok, DeepSeek) for intelligent responses.
• Contextual Responses: Upload CSVs or provide website context for accurate, relevant answers.
• User Authentication: Integrated with Clerk for secure user management.
• Dashboard & Analytics: Track bot usage, ratings, and chat history.
• Rate Limiting & Security: Protects APIs with IP and API key rate limiters.
• Modern Stack: Built with React, Tailwind CSS, Node.js, Express, and MongoDB.

---

Tech Stack

• Frontend: React.js, Tailwind CSS, Clerk (authentication)
• Backend: Node.js, Express.js, MongoDB (Mongoose)
• AI/ML: Google Gemini, OpenRouter, Xenova Transformers
• NPM Package: botly-bot for easy React integration

---