Problem: AI breaks character
Solution: Reinforce the role mid-conversation: "Remember, you're responding as [role] who believes [philosophy]"Problem: Responses feel generic despite role
Solution: Add more specific background and examples of their work/approachProblem: Role is too limiting
Solution: Add versatility: "While your specialty is X, you also have working knowledge of Y, and Z"Problem: Multiple roles create confusion
Solution: Clearly separate responses: "Speaking as [Role 1]:... Now as [Role 2]:..."