llm

Overview ¶

Package main demonstrates LLM integration with Romancy workflows.

This example shows:

• Ad-hoc LLM calls with llm.Call()
• Structured output with llm.CallParse()
• Multi-turn conversations with llm.CallMessages()
• Reusable LLM definitions with llm.DefineDurableCall()
• App-level LLM defaults with llm.SetAppDefaults()

The key benefit is that all LLM calls are automatically cached as activities, so workflow replay will return cached results without re-invoking the LLM API.