Comment by ivanovm

2 days ago

would just caching llm responses work here?

Which responses? For how long? With what level of detail? Those are the questions we are all trying to figure out now, and the performance of your agent is highly dependent on the answer to that.