Adding Google AI | #12 Building an App with Effect

👉 Enroll to Practical Effect here: https://lucasbarake.com/ Buy me a coffee! https://buymeacoffee.com/lucasbarake Repo: https://github.com/lucas-barake/building-an-app-with-effect This video picks up where the last one left off, starting with a fix to make our AI service compatible with more advanced reasoning models by improving how we handle their complex responses. The main focus is expanding our service to support multiple AI providers, and we walk through adding a complete integration for Google's AI. During this process, we dive into an advanced security pattern for safely managing API keys. To finish, we refactor the service architecture to be truly provider-agnostic, making it easy to switch between different AI models in our application. 00:00:00 - Fixing the "Reasoning Models" Flaw 00:02:06 - Integrating Google AI: The Client Setup 00:04:01 - The Security Problem: Leaking API Keys in Traces 00:04:24 - The Solution: FiberRef & Scoped Header Redaction 00:06:17 - Explainer: How FiberRef Works (An Analogy) 00:08:20 - Implementing the Google AI Model Methods 00:09:28 - Handling Google's Strict Schema Requirements 00:10:33 - Testing the Google AI Integration 00:11:23 - Refactoring for a Provider-Agnostic Usage 00:12:32 - Converting Concrete Models to Abstract Layers 00:14:04 - Why Chaining Effect.provide is Discouraged 00:14:59 - Final Refactor: Renaming Providers to Clients 00:15:46 - Next Steps: Building the RPC Streaming Layer