· Michael Antensteiner · Product  · 2 min read

Introducing the Lyte Stack

A self-hosted, batteries-included stack for running modern applications — Lytebase, Lytegate, Lytewatch, and Lytebulb. Four core services designed to work together while remaining usable independently.

A self-hosted, batteries-included stack for running modern applications — Lytebase, Lytegate, Lytewatch, and Lytebulb. Four core services designed to work together while remaining usable independently.

We have been building Lytebase, a self-hostable application platform based on K3s. Running our own workloads on it has surfaced recurring needs across the applications we ship:

  • A common place for service connections, background jobs, and policy instead of rebuilding integration glue per app
  • An operations console for first-response diagnosis without assembling a full observability stack
  • An AI layer for model routing and reviewable agent work

Instead of reimplementing these in every application, we made them shared platform products. The result is the Lyte Stack — four self-hosted products designed to work together while remaining usable independently. The stack is not meant to be closed: Lytebase should make the Lyte services easy to run together, while still letting teams bring alternatives or any other container-based service where they fit better.

The four core services

  • Lytebase — The foundation. A deterministic CLI and control-plane API for deploying and operating applications on K3s, with managed platform services.
  • Lytegate — Governance. One place for tenant-scoped service connections, worker-backed actions, jobs, and policy — shared across applications instead of rebuilt each time.
  • Lytewatch — Operations. A first-response console that brings health, logs, metrics, and incident events into one self-hosted UI. Built for diagnosis, not heavyweight observability.
  • Lytebulb — AI. An intelligence gateway for routing LLM requests across configured providers, plus workspace agents that operate in real Git repositories.

All four are self-hosted, built on open-source foundations, and designed for teams that want control over their infrastructure. Each can run standalone — adopt any combination, or use all four as the full stack.

Where we are

  • Lytebase is running our own production workloads.
  • Lytegate is being built around governed integrations and worker-backed jobs, and Lytewatch already uses it to deliver alerts through notification channels.
  • Lytewatch has its operational breadth implemented and is undergoing stability validation.
  • Lytebulb ships the intelligence gateway with provider routing, retained threads, and workspace agents.

ThinXdone, our multi-organization project management platform with smart workspace integration, runs entirely on the Lyte Stack and helps validate that the approach works.

What comes next

More writing will follow as products stabilize and pilots wrap up.

Back to Blog

Related Posts

View All Posts »
Building Lytebase - A Leaner Path to Cloud-Native

Building Lytebase - A Leaner Path to Cloud-Native

Kubernetes has become the de facto engine of modern software. Its elasticity, strong ecosystem, and proven reliability are hard to ignore—whether you run a SaaS in the cloud or deliver on-prem solutions.

Welcome to AppSpark!

Welcome to AppSpark!

Humane Software for the web. We want to create useful web tools that are fair in regards to the people that use them. We love the internet, but the current application-model is broken, because it is not honest.