Introducing Genkit for Go: Build Generative AI Apps with Native Go

Google has launched Genkit for Go, an open‑source framework that enables developers to build generative AI applications and cloud services natively in Go. The project is a collaboration between the Google Firebase and Go teams and entered alpha on July 17.

Genkit for Go lets Go’s performance and concurrency be combined with Genkit’s libraries and tools to create AI assistants, retrieval‑augmented generation (RAG) agents, and data‑conversion utilities that turn unstructured text into structured formats such as SQL queries.

It follows the earlier release of Firebase Genkit for Node.js (May 2023) and provides lightweight, composable abstractions that simplify complex AI workflows without sacrificing control or customisation.

Key features

Unified generative API for accessing large language models like Gemini or Gemma through a single interface.

Built‑in observability and debugging for AI workflows.

Native vector‑database support, enabling RAG with cross‑provider indexing and retrieval APIs.

Core components

Ready‑to‑use Go libraries that follow standard Go conventions and can be dropped into existing projects.

Dotprompt file format for simplified prompt engineering, allowing templates, patterns, and model configurations in a single .prompt file.

Plugin system offering integrations with Google AI, Vertex AI, Ollama, Pinecone and other LLM and vector‑db providers.

Developer tools including a CLI and a browser‑based UI for iterating AI workflows, observing performance, and evaluating models.

Production observability through integration with Google Cloud Operations suite and OpenTelemetry.

Developers can start with the official getting‑started guide on Firebase, use VS Code or Google’s Project IDX cloud IDE to open the Genkit UI, and submit issues or feature requests on GitHub. A Discord server is also available for community support.