EmbedMyBot - No-Code AI Chatbot Platform

What is EmbedMyBot - No-Code AI Chatbot Platform?

EmbedMyBot is a no-code AI chatbot SaaS I built to help businesses train chatbots on their own content (documents + website), deploy them to any website via a one-line embed, and capture leads. It uses a RAG pipeline for grounded, cited answers, Stripe for subscription billing, and real-time analytics, with an embeddable widget, a widget customizer, and a full admin console.

Technologies & Tools Used

Technical Highlights

• Next.js - App Router architecture, Server Components + SSR/SSG
• React - Component-driven UI, Hooks & state management
• TypeScript - Strict typing, Scalable data models
• TailwindCSS - Utility-first styling, Responsive + dark/light theming
• Supabase - Postgres database, Auth & Row-Level Security, Realtime + pgvector
• AI / RAG - Retrieval-augmented generation, Citation-backed answers
• LLM Integration - Google Gemini (embeddings + vision), DeepSeek (chat generation)
• Vercel AI SDK - Streaming chat responses, Unified model interface
• pgvector - Vector similarity search, Semantic knowledge retrieval
• Web Crawler - robots.txt + sitemap aware, HTML extraction (cheerio + turndown)
• Embeddable Widget - One-line embed script, Sandboxed iframe + postMessage API
• recharts - Interactive analytics charts, Usage & CSAT visualizations
• lucide-react - Consistent icon system, Clean UI visuals
• Stripe - Monthly/yearly subscriptions, Webhooks, proration & Customer Portal
• SEO / AEO - JSON-LD structured data, Dynamic sitemap, robots & llms.txt

Key Features

• Designed and built a multi-tenant AI chatbot SaaS using Next.js (App Router), TypeScript, and Supabase.
• Engineered a Retrieval-Augmented Generation (RAG) pipeline with vector embeddings (pgvector) to deliver grounded, citation-backed answers and prevent hallucinations.
• Developed a web crawler and document ingestion system (PDF, DOCX, Markdown) that respects robots.txt and sitemaps, with automatic chunking, embedding, and indexing.
• Built an embeddable JavaScript widget (async loader + sandboxed iframe) with a one-line embed, a customizable UI, and a postMessage JavaScript API for third-party sites.
• Created a no-code widget customizer with themes, brand colors, chat background patterns, avatars, and domain allowlisting.
• Integrated Stripe subscription billing with monthly/yearly plans, proration, webhooks, the Customer Portal, and anniversary-based renewal cycles.
• Implemented server-side, plan-based usage enforcement for messages, image messages, chatbots, and storage.
• Built analytics dashboards (conversation volume, response latency, CSAT feedback, unanswered-question tracking) and a built-in leads CRM with CSV export.
• Added vision support (image messages), multilingual replies (90+ languages), and a public demo bot that carries its configuration into a new account on signup.
• Developed an internal admin console for user, usage, and revenue management, plus a marketing site with a blog, docs, and dynamic SEO landing pages.
• Optimized for SEO and AI answer engines with structured data (JSON-LD), a dynamic sitemap, AI-crawler robots rules, llms.txt, and per-page Open Graph cards.

Project Category