Skip to main content
Version: 3.20.0 (stable)

Platform Guide

The Platform Guide provides comprehensive documentation for developers building mims on mimOE. This is Entry Point B, for developers who want full control over the runtime environment and platform capabilities.

Who This Guide Is For​

This guide is for developers who need to:

  • Build custom mims with JavaScript or WebAssembly
  • Understand the mimOE runtime architecture
  • Deploy applications directly using the MCM API
  • Work with platform-level APIs and design trade-offs
  • Optimize performance for device-first execution
Looking for an easier start?

If you want to run AI models on-device with minimal friction, consider starting with the AI Foundation Package instead. You can always return to this guide when you need deeper platform integration.

Two Development Approaches​

mimOE supports two approaches for building mims and AI agents:

JavaScript Development​

Build mims using JavaScript (ES5-compatible via the mimOE JavaScript runtime). This is the recommended approach for most developers.

Best for:

  • Rapid development
  • AI agent logic and orchestration
  • Web developers familiar with JavaScript
  • Most application code

Learn more: JavaScript Development

WebAssembly (WASM) Development​

Build high-performance mims using Rust, C, or C++ compiled to WebAssembly.

Best for:

  • Performance-critical workloads
  • Porting existing native code
  • Advanced systems programming
  • Specialized algorithms

Learn more: WASM Development

Platform Guide Sections​

Architecture​

Understand how mimOE works under the hood:

Getting Started​

Set up your development environment:

JavaScript Development​

Build mims with JavaScript:

WASM Development​

Build high-performance mims with WebAssembly (coming soon):

  • When to Use WASM: Deciding between JavaScript and WASM
  • Rust Toolchain: Set up Rust/C development for WASM
  • WASM mim Runtime API: Available runtime functions

Development Workflow​

Regardless of whether you choose JavaScript or WASM, the development workflow follows the same pattern:

  1. Write Code: Develop your mim using JavaScript or WASM
  2. Build/Bundle: Transpile and package for mimOE runtime
  3. Deploy: Use MCM API to deploy your mim
  4. Test: Verify functionality on target platform
  5. Iterate: Refine and improve

Key Concepts​

Serverless-on-Device​

mimOE implements serverless execution on edge devices. mims are instantiated on-demand per request and terminated after response, minimizing memory footprint.

Key implications:

  • No state persists between requests
  • Each request starts with a fresh instance
  • Use provided storage APIs for persistence

Device-First Continuum (DFC)​

Processing starts on the device where data originates. Cloud services are available when needed, but not required.

Benefits:

  • Reduced latency
  • Enhanced privacy
  • Offline capability
  • Cost efficiency

Request-Driven Execution​

All mims are triggered by HTTP requests. There are no long-running processes or background tasks.

What this means:

  • mims are REST API endpoints
  • Execution is synchronous (request → response)
  • For async patterns, use storage and polling

API Reference​

For detailed API documentation, see:

Next Steps​

Choose your path based on your needs:

Entry Point A (Easier Path)

If you haven't already, consider trying the AI Foundation Package first. It provides a gentler introduction to mimOE with zero-code AI inference, and you can progress to custom development when ready.