ChatEaver
AI Customer Support & RAG Chatbot Platform — Embeddable Widget, Docs & Editorial Blog.
The Problem
Businesses need an AI customer support assistant that answers visitor questions using verified business knowledge without hallucinations, while offering a full-featured documentation center, editorial blog, and embeddable website chat widget.
The Approach
Built the complete web UI for ChatEaver using Next.js 16 App Router, React 19, TypeScript, and Tailwind CSS v4 (Warm Ember design system). Developed a responsive documentation system (/docs and in-app /dashboard/docs) with fluid multi-column grids, automated export/import scripts, an SEO-first editorial blog (/blog) with TipTap rich text rendering, and a standalone Vite IIFE embeddable website widget (/sdk/chat-widget.js).
Key Challenges
Achieving zero-side-effect isolation for the embeddable chat widget script across third-party websites required a standalone Vite IIFE bundle with custom state management. Responsive docs layout needed custom CSS grid breakpoint logic to maintain flush 0px padding inside the dashboard shell while keeping 16px padding on public docs.
Technical Decisions
Next.js 16
App Router with React 19 Server Components for high-performance SEO docs, blog pages, and BFF proxying to FastAPI backend.
Tailwind CSS v4
CSS-first Warm Ember design system with custom variable tokens, dark mode, and responsive layout utilities.
Vite (IIFE Widget)
Standalone lightweight script bundle (/sdk/chat-widget.js) embeddable via single script tag on third-party sites.
TipTap + HTML
Rich text content rendering for blog articles and documentation guides with code blocks and custom callouts.
The Result
Shipped a production-ready Next.js 16 + React 19 frontend with 330+ passing automated tests, custom AI studio builder, Knowledge Base error wrapping, and seamless automated database seed/export workflows.
Next project
MijuDrama