Xpell 2 — AI-Native Architecture for Vibe Coding

AI can generate code. But generation alone is not architecture. If AI is going to collaborate with live systems — update interfaces, evolve flows, and persist state — the runtime must be designed for it.

At Aime Technologies, we call this AI-Native Architecture: an approach where software is structured to support real-time collaboration between humans and AI.

What “AI-Native Architecture” means

  • Interfaces are structured data — not static code that must be rebuilt for every change.
  • Logic can be serialized — enabling safe AI-driven edits without rewriting everything.
  • State is formalized — predictable shared runtime state instead of scattered ad-hoc variables.
  • Runtime interpretation replaces rebuild cycles — evolve the application while it runs.
  • AI can collaborate safely — modify parts of the system without breaking the whole.

Xpell 2 is a full-stack platform

Xpell 1 focused on the client runtime. Xpell 2 expands into a structured, modular, full-stack platform.

@xpell/core

  • XData 2 — improved shared runtime state model.
  • Nano-Commands 2 — command handlers as text and structured JSON.
  • XEM — Xpell Event Manager.
  • XModule — extension point for runtime behavior.
  • XObject — core runtime object model.

@xpell/node (Xpell server / xnode)

  • Xpell Server — server execution layer.
  • Wormholes 2 — real-time client/server bridge.
  • XDB — Xpell database (graph + vector / semantic-ready).

@xpell/ui (Web)

  • XUI — real-time UI framework.
  • XUIObject — UI object model (UI behaviors live here, not in core).
  • Wormholes (client) — real-time connectivity.

@xpell/3d

  • Spatial runtime layer based on three.js.

Designed for Vibe Coding

Xpell 2 is designed for Vibe Coding — where AI can generate and evolve applications through structured runtime contracts rather than static code output.

Before launching the full Vibe experience, we will share an early Codex skill capable of generating complete Xpell applications on top of Xpell 2 Alpha.

Technical documentation, examples, and downloads will live on xpell.ai.

Explore Xpell.AI