Absolutely,
I have been testing how we can auto add highlights and tags, and then create magic insights bases on that. The auto highlights worked great, but when I went to create magic insights based on titles, it only looked at one of the two interviews in the project, so it only got half of the information. It also reused the same highlights and wrote almost the same thing across almost all the insights. This told me that it wasn´t very thorough in the creation of magic insights, so I asked contextual chat why, and it confirmed that the insight seemed like it only looked at one note, and that it wasn´t very thorough.
I then tried getting chat to suggest insight titles that would dig deeper, but it seems like the same error occurred. Also, the insights are all almost the same length, which seems weird. The URL is here for the insights page (the bucket called Nanna is the one we made first, the other one is the ones chat suggested) https://grundfos-insights.dovetail.com/projects/4qrYiILW8ZlqyIGsZMC6ye/v/6FXnWg32eDLWC7CD8Zi9tV