Skip to main content
webAI isn’t just an application — it’s a platform. You can build your own tools, utilities, and collaborative experiences that run as first-class apps inside the webAI shell. Your apps get access to the same APIs that power the built-in apps: local AI inference, peer-to-peer collaboration, user identity, end-to-end encryption, and navigation. No server required. No API keys. Everything runs on-device.

​
What you can build

AI-powered tools

Build apps that run AI inference locally — summarizers, writing assistants, code generators, analyzers — all private, all offline-capable.

Collaborative apps

Create real-time multiplayer experiences. The platform handles networking, state sync, and conflict resolution via .

Standalone utilities

Dashboards, calculators, games, note-taking tools — anything that runs as a web page can run as a webAI app.

​
How it works

Apps are single HTML files that run inside a sandboxed iframe within the webAI shell. The shell injects a set of JavaScript globals into your app’s environment, giving you access to platform capabilities without importing any libraries.
webAI Shell (Apogee)
Your App (iframe)
window.OasisHost → AIwindow.ApogeeShell → Navwindow.CollaborationManager → P2Pwindow.UserIdentityManager → Authwindow.E2ECrypto → Encryption

​
Platform APIs at a glance

APIPurposeKey capabilities
OasisHostOn-device AI inferenceAcquire models, stream completions, check runtime status
CollaborationManagerPeer-to-peer collaborationHost/join spaces, sync state with CRDTs, real-time chat
UserIdentityManagerUser identity and authGet device identity (), display names, auth headers
ApogeeShellShell navigationSwitch views, return to launcher, route between apps
E2ECryptoEnd-to-end encryptionEncrypt and decrypt data between peers

​
Start building

1

Understand the app model

Learn how apps run inside the webAI shell, what constraints apply, and how to structure your project.App architecture →
2

Access platform APIs

Learn the universal pattern for safely accessing shell-injected globals from your app code.Accessing shell APIs →
3

Add AI to your app

Connect to the on-device Oasis AI runtime to stream completions, check model status, and build intelligent features.OasisHost API →
4

Build, bundle, and deploy

Package your app as a single HTML file and upload it to the webAI launcher.App lifecycle →
You don’t need to use every API. Most apps start as standalone tools using just the AI runtime. Add collaboration, identity, or encryption as your app grows.