I often use Claude as a discussion partner when I have a moment. My favorite mode is to use it in speech mode, so I can be out on a walk and have a conversation with the model about something that I am thinking about. This works really well and helps me structure my thoughts, come up with new ideas and plan out the things I should focus on.
The only challenge is that Claude tends to be very positive and supportive and not as critical as I would like. I noticed it with the founder of a startup I am involved in where the individual approached the sales process way too naive with the assumption that a few posts on LinkedIn would fill the sales funnel without concerns. Because Claude had said so.
It can feel soothing and you feel like a million dollars while talking to the model, but when reality hits, it rapidly becomes clear that things are not as easy and smooth as you might have imagined. So, my instructions to Claude are to be much more critical and to take less of a “feel good” role. As a reflection for today: How do you use AI and make sure that you get realistic feedback?