How this works

1. Pick a case from the list on the left.
2. Read the conversation, then compare side by side: what the AI did on the left, the correct answer on the right.
3. Decide: approve it, fix it and save, or reject it with a note.

The colored result only says whether the suggestion agrees with what the AI did — different is often a deliberate correction. You decide who is right.