About Tower

An AI-powered assistant for composable website technology with emphasis on Contentstack. Tower uses RAG (Retrieval Augmented Generation) to provide informed responses from a curated knowledge base.

Features

AI Chat

Multi-model chat interface supporting Claude Sonnet/Opus, ChatGPT, and Kimi. RAG-powered responses draw from the knowledge base. Includes conversation history and tagging for organization.

Open Chat →

Stories

Create and manage custom scenarios for different use cases. Organize stories with tags for easy discovery and reuse.

Browse Stories →

Knowledge Base

Document ingestion from text and URLs. Content is chunked and converted to vector embeddings for semantic search, powering the RAG system.

Manage Documents →

Library

Personal file storage with folder organization. Upload, manage, and share files within the platform.

Open Library →

Interactive Simulators

Learn Contentstack concepts through hands-on simulation tools.

Tech Stack

Frontend

  • Next.js 16 with App Router
  • React 19 for UI components
  • Tailwind CSS 4 with dark mode
  • Vercel AI SDK v4 for streaming
  • Mermaid for diagrams

Backend

  • Hono.js lightweight API framework
  • PostgreSQL with pgvectorscale (1536-dim vectors)
  • Drizzle ORM for type-safe queries
  • Redis for caching
  • S3-compatible storage for files

AI / ML

  • OpenRouter API (Claude 3.5 Sonnet default)
  • OpenAI text-embedding-3-small for embeddings
  • RAG pipeline: ingest → chunk → embed → retrieve → augment

Authentication

  • Google OAuth via Auth.js
  • Database-backed sessions for security

Hosting

Tower is hosted on Railway, with separate services for the frontend and backend. The platform provides:

  • PostgreSQL with pgvectorscale extension
  • Redis for session and data caching
  • S3-compatible object storage
  • Automatic deployments from Git

Admin Tools