genkit

Genkit is an open-source framework from Google's Firebase team for building applications that use AI models. It provides programming libraries (SDKs) for JavaScript, TypeScript, Go, Python, and Dart, allowing developers to add AI features to a project in whichever language they are already using. The JavaScript and Go versions are production-ready; the Python version is in beta; and the Dart version is an early preview.

The framework's main idea is to give you a single consistent way to work with many different AI providers. Instead of learning a separate API for Google Gemini, OpenAI, Anthropic, or a locally running model via Ollama, you write code once using Genkit's API and swap the underlying model by changing a configuration line. This covers text generation, image generation, structured data output, tool calling (where the AI can invoke functions in your code), and retrieval-augmented generation (where the AI pulls relevant documents before answering).

Genkit ships with a local developer UI and command-line tool that let you test prompts interactively, compare outputs from different models side by side, and step through detailed execution traces to debug problems. You can iterate on prompts and AI workflows without deploying to a server.

For deployment, Genkit is designed to run on the server side. You can deploy to Firebase's Cloud Functions, Google Cloud Run, or any other environment that supports your chosen language. Client SDKs and helpers are provided for connecting web and mobile frontends (Next.js, React, Angular, iOS, Android) to the server-side AI logic.

In production, the framework includes monitoring tools for tracking request volumes, latency, and error rates so you can catch problems in live usage. The documentation and a set of interactive example applications are available at genkit.dev.