Our Technology

Modern, scalable cloud infrastructure designed for millions of requests with sub-second response times.

Every surface, every signal, on one page.

How the platform actually fits together — from the surfaces a reader touches, through the engines that translate, narrate, and answer, down to the storage and retrieval that hold it all up. Hand-drawn, not a logo dump.

Overview diagram of the kaOS architecture, showing surfaces, engines, retrieval, storage, and operations as a single composed system.
kaOS system architecture, one frame.

Thirteen frames, each a different cut.

Click any frame to see the system from a particular angle. Same architecture, drawn in the language of the question you came with.

Tech stack

Frontend

  • Next.js
  • React
  • Tailwind CSS
  • TypeScript

Backend & APIs

  • Node.js
  • Supabase PostgreSQL
  • pgvector
  • Cloud Run

AI & ML

  • Anthropic Claude
  • Google Gemini
  • Cohere
  • Camb.ai

Infrastructure

  • Vercel
  • Google Cloud
  • Supabase Storage
  • GitHub Actions

Key capabilities

Semantic Search

Vector embeddings with pgvector across our corpus. Not keyword matching — actual understanding.

Translation Pipeline

23 languages. Professional quality with publishing metadata preservation.

Text-to-Speech

Multiple voice providers, character-aware pacing, narrative understanding.

Content Generation

Bookcasts, Sparks, summaries, analysis — trained on literary analysis.

Self-Healing Infrastructure

Recursive agent pipeline monitors, analyzes, and fixes issues autonomously.

MCP Integration

Model Context Protocol server for AI agent access to all kaOS capabilities.