What problem does it solve?
Adding AI capabilities like LLM chat, embeddings, or web search to a Golem agent normally requires learning each provider's API. This Skill provides unified, provider-agnostic Rust crates so you can integrate AI features with a single consistent API and swap providers by changing one dependency.
Core Features & Use Cases
- Unified AI APIs: Core crates for LLM chat, embeddings, web search, document search, graph/vector databases, video generation, speech-to-text, and text-to-speech, each with multiple provider backends (OpenAI, Anthropic, Bedrock, Qdrant, Neo4j, ElevenLabs, and more).
- Durable Execution: Durable* provider types record responses in Golem's oplog and replay them on recovery, making AI calls fault-tolerant.
- Multi-turn Conversations: Maintain event history as agent state for chat sessions with tool calling and streaming support.
- Use Case: Build a chat agent mounted at /chats/{chat_name} that answers questions via OpenAI, then switch to Anthropic later by changing only the type alias and Cargo dependency.
Quick Start
Add LLM chat to my Rust Golem agent using the golem-ai-llm crate with OpenAI as the provider.