Wrong answers

Agents can mark if an answer given by the AI is wrong or partially correct. This way it is easier to recognize when the AI did not answer correctly. On the other hand, it allows to evaluate by chat, what % of the answers given by the AI are bad, regular and good (which would be the not bad and not regular ones).