Voice SDK promises to code your iOS app while you talk to it

What it does

This is the iOS client SDK for the Alan AI platform. You embed it into a Swift or Objective-C app, add a voice button, and users can talk to an AI agent. The agent is scripted in JavaScript via Alan AI Studio, a separate web IDE. The SDK handles speech recognition, streaming audio to Alan’s cloud, and executing the returned commands in your app.

The interesting bit

The pitch is “Application-Level AI” — the voice agent doesn’t just parse intent, it allegedly generates business logic and UI at runtime using your app’s APIs, docs, and GUI structure as context. That’s a much bigger claim than typical voice SDKs, though the README offers no technical detail on how this “Three-Layer AI” actually works or what guardrails exist.

Key highlights

• Supports both Swift and Objective-C iOS apps
• Dialog scripts are written in JavaScript, edited and tested in Alan AI Studio
• Distributed via CocoaPods and GitHub releases
• Includes example apps (Food Delivery demo with voice commands like “What do you have?”)
• Cross-platform: sibling SDKs exist for Android, Flutter, React Native, Web, Ionic, Cordova, and PowerApps

Caveats

• The README contains a copy-paste error: it says the iOS SDK “enables you to embed Alan’s intelligent layer into your Android applications”
• No code, architecture diagrams, or implementation details visible in the repository — the SDK appears to be a closed-source binary framework
• “Self-coding” and “no developers needed” claims are aspirational marketing; no evidence or benchmarks provided in the sources

Verdict

Worth a look if you’re building voice-driven iOS apps and want a managed, cross-platform voice AI backend with a scriptable dialog layer. Skip it if you need transparency into the model, on-device inference, or are skeptical of black-box “AI generates your code” promises without proof.