Xpell.ai logo
Xpell Xpell 2 Beta launch site
Xpell 2 Beta

Real-time Vibe Coding — No rebuild. No wasted tokens.

Xpell 2 is an AI-native runtime that lets you modify your app while it's running. Generate, preview, and apply changes instantly — without recompiling your code.

Build dashboards, internal tools, and AI interfaces in minutes — without rebuild loops or full code generation.

Get Started in 2 Minutes See it in action
Modify your app live. Preview changes instantly. Save only what you accept.

Xpell 2 Beta

Run the app, prompt the runtime, preview the result, then persist the change. No rebuild loop required.

Prompt Describe the change you want in the running system.
Preview Inspect the generated UI or behavior before saving it.
Apply Accept the runtime change without recompiling the app.
Persist Save the accepted structure back into the project.

2-Minute Start

Start from the runtime starter, boot the local app, then modify it while it runs.

Install and run

git clone https://github.com/xpell-ai/xpell-vibe-starter
pnpm install
pnpm dev

What happens next

Open IntelliSphere (the in-app AI editor) → type → UI changes instantly.

Starter repo: xpell-ai/xpell-vibe-starter

Modify your app live

Xpell keeps the app running while changes are generated, previewed, and applied in real time.

Watch how a running app is modified live — no rebuild, no reload.

Using IntelliSphere

IntelliSphere is the in-app AI editor that lets you edit the running system in real time. Type a prompt, preview the result, and apply changes instantly.

  1. RunStart the app and keep the runtime active.
  2. PromptDescribe the change you want in plain language.
  3. UI updatePreview the generated interface update immediately.
  4. ApplyAccept the change when the runtime output is correct.
  5. SavedPersist the accepted structure back into the project.

Example prompt

Add email field to users

Try this in the starter app → xpell-vibe-starter

Use the running system as the editing surface instead of regenerating a full application snapshot.

Videos

Recorded walkthroughs for the runtime workflow, focused on live edits inside the running app.

Xpell 2 Real-time Demo

Run the starter app, prompt changes, and apply them live without rebuilding.

Stop regenerating apps. Start evolving them.

Xpell treats AI as part of a running interpreter instead of a one-shot generator.

Traditional AI builders Xpell
Regenerate full app output after each prompt. Modify the running app directly.
Rebuild or redeploy to inspect every change. Preview changes without rebuild loops.
Spend tokens repeating unchanged structure. Spend tokens on the delta you actually need.
Prompt output sits outside runtime state. Prompt output becomes part of runtime state and can be persisted.

Architecture

Xpell is a real-time interpreter, not a framework.

@xpell/core

Runtime engine primitives, state, events, and deterministic contracts.

@xpell/node

xnode runtime for server execution, modules, and runtime operations.

@xpell/ui

Runtime-driven UI layer that updates while the app is running.

@xpell/3d

3D runtime layer using the same interpreter model for spatial apps.

AI that edits your app while it runs

Prompt -> runtime -> persist. Generate the change, preview it in the running app, then save the accepted result into the project.

Getting Started

Move from the quick starter into the step-by-step Xpell 2 Beta flow.

/getting-started/

Examples

Inspect focused runtime examples after the starter is running locally.

/examples/

GitHub

Use the starter repo and package sources as the primary code entry point.

github.com/xpell-ai

Videos

Watch the published runtime-editing walkthroughs when they are available.

#videos