← Back to context

Comment by Nuzzerino

19 hours ago

> We need llms to be able to tap that not add the same functionality a layer above and MUCH less efficiently.

Agents, tool-integrated reasoning, even chain of thought (limited, for some math) can address this.