1. create a goroutine 2. call the llm api in a loop 3. if you get back a tool invocation then do the tool 4. if you get back the answer send it to the caller using a channel since this style is so good for ai agent actions we could give it a name, like “actor model”
Jo Kristian Bergum
Jo Kristian Bergum8.8. klo 13.30
We need more 200-lines examples of building an agent loop with tools using the primitive APIs. We need to demystify agent implementations. We don’t need another agentic framework.
@0xz80 hello joe
henry 🌘
henry 🌘18 tuntia sitten
waiting for a major lab to drop an announcement video that goes this hard
4,67K