Comment by drew-y

How does automation reasoning actually check a response against the set of rules without using ML? Wouldn't it still need a language model to compare the response to the rule?