How they work together
GenerableWorker gives your agent a secure Cloudflare backend to manage sessions and spending limits. Using GenerableJev, your agent can set up your app to get low-cost, typed answers and confidence scores through your Worker. For advanced text capabilities, your agent can implement GenerableLLM or GenerableAWS so your app can access chat models through Apple's FoundationModels APIs.
Together, these tools let your agent add intelligence to your Swift apps without ever shipping API keys. You can start with Apple's Private Cloud Compute models and configure external models as a fallback when Apple's models fail. For simpler tasks, use Jev to keep costs low and only call more advanced models when needed. Since everything uses the familiar FoundationModels APIs, your agent can work with a consistent Swift interface and test with offline stand-ins without making paid model calls.
You'll need Xcode 27, an app targeting iOS 27 or macOS 27, and a Cloudflare account with AI Gateway credits. GenerableAWS also requires an AWS account with Amazon Bedrock in US East (Ohio).
What’s inside

GenerableWorker
Use lower-cost AI models in your Swift apps through Cloudflare Workers. No client-side API keys. Secure with App Attest, built for production.

GenerableJev
Easily use TypeSafe's new low-cost Jev decision model in your Swift apps. No client-side API keys. Secure Cloudflare backend with App Attest protection. Built for production.

GenerableLLM
GenerableLLM lets you run remote AI models from OpenAI, Anthropic, and Cloudflare using Apple's FoundationModels SDK. Your Swift app connects through a secure Cloudflare backend without exposing API keys. Switch models, control costs, and keep using the familiar FoundationModels APIs.

GenerableAWS
Use Amazon Bedrock models via Apple's FoundationModels SDK. No client-side API credentials. Secure with App Attest and production-ready.