ruby_llm

Why we included this project

Every major AI provider ships its own Ruby client with different APIs and response formats, so supporting more than one is normally a mess of adapter code. RubyLLM replaces that with a single set of methods that work the same across OpenAI, Anthropic, Google, xAI, AWS, and local Ollama models, letting you swap providers without rewriting your application. On top of plain chat it handles tool calling, agents, structured JSON output, embeddings, image generation, and audio transcription and speech, all in short readable calls that fit naturally into a Rails codebase. That covers a quick chatbot prototype as well as heavier jobs like pulling text out of PDFs and images or a RAG pipeline that needs several model capabilities at once. It is a gem you run yourself rather than a hosted service, so you bring your own API keys and run it wherever your Ruby app already lives; Chat with Work, a fully private work AI, uses it in production.